id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
list |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2507.02976
|
Amirali Sajadi
|
Amirali Sajadi, Kostadin Damevski, Preetha Chatterjee
|
Are AI-Generated Fixes Secure? Analyzing LLM and Agent Patches on SWE-bench
| null | null | null | null |
cs.CR cs.LG cs.SE
|
http://creativecommons.org/licenses/by/4.0/
|
Large Language Models (LLMs) and their agentic frameworks are increasingly adopted to automate software development tasks such as issue resolution and program repair. While prior work has identified security risks in LLM-generated code, most evaluations have focused on synthetic or isolated settings, leaving open questions about the security of these systems in real-world development contexts. In this study, we present the first large-scale security analysis of LLM-generated patches using 20,000+ issues from the SWE-bench dataset. We evaluate patches produced by a standalone LLM (Llama 3.3) and compare them to developer-written patches. We also assess the security of patches generated by three top-performing agentic frameworks (OpenHands, AutoCodeRover, HoneyComb) on a subset of our data. Finally, we analyze a wide range of code, issue, and project-level factors to understand the conditions under which LLMs and agents are most likely to generate insecure code. Our findings reveal that the standalone LLM introduces nearly 9x more new vulnerabilities than developers, with many of these exhibiting unique patterns not found in developers' code. Agentic workflows also generate a significant number of vulnerabilities, particularly when granting LLMs more autonomy, potentially increasing the likelihood of misinterpreting project context or task requirements. We find that vulnerabilities are more likely to occur in LLM patches associated with a higher number of files, more lines of generated code, and GitHub issues that lack specific code snippets or information about the expected code behavior and steps to reproduce. These results suggest that contextual factors play a critical role in the security of the generated code and point toward the need for proactive risk assessment methods that account for both code and issue-level information to complement existing vulnerability detection tools.
|
[
{
"version": "v1",
"created": "Mon, 30 Jun 2025 21:10:19 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:50:13 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Sajadi",
"Amirali",
""
],
[
"Damevski",
"Kostadin",
""
],
[
"Chatterjee",
"Preetha",
""
]
] |
2507.02987
|
Sonia Laguna
|
Andrea Agostini, Sonia Laguna, Alain Ryser, Samuel Ruiperez-Campillo, Moritz Vandenhirtz, Nicolas Deperrois, Farhad Nooralahzadeh, Michael Krauthammer, Thomas M. Sutter, Julia E. Vogt
|
Leveraging the Structure of Medical Data for Improved Representation Learning
| null |
Published at the ICML 2025 Workshop on Multi-modal Foundation Models and Large Language Models for Life Sciences
| null | null |
cs.CV cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Building generalizable medical AI systems requires pretraining strategies that are data-efficient and domain-aware. Unlike internet-scale corpora, clinical datasets such as MIMIC-CXR offer limited image counts and scarce annotations, but exhibit rich internal structure through multi-view imaging. We propose a self-supervised framework that leverages the inherent structure of medical datasets. Specifically, we treat paired chest X-rays (i.e., frontal and lateral views) as natural positive pairs, learning to reconstruct each view from sparse patches while aligning their latent embeddings. Our method requires no textual supervision and produces informative representations. Evaluated on MIMIC-CXR, we show strong performance compared to supervised objectives and baselines being trained without leveraging structure. This work provides a lightweight, modality-agnostic blueprint for domain-specific pretraining where data is structured but scarce
|
[
{
"version": "v1",
"created": "Tue, 01 Jul 2025 11:14:45 GMT"
},
{
"version": "v2",
"created": "Wed, 09 Jul 2025 19:45:03 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 12:44:31 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Agostini",
"Andrea",
""
],
[
"Laguna",
"Sonia",
""
],
[
"Ryser",
"Alain",
""
],
[
"Ruiperez-Campillo",
"Samuel",
""
],
[
"Vandenhirtz",
"Moritz",
""
],
[
"Deperrois",
"Nicolas",
""
],
[
"Nooralahzadeh",
"Farhad",
""
],
[
"Krauthammer",
"Michael",
""
],
[
"Sutter",
"Thomas M.",
""
],
[
"Vogt",
"Julia E.",
""
]
] |
2507.03476
|
Toshiyuki Kobayashi
|
Kazuki Kannaka, Toshiyuki Kobayashi
|
Zariski-dense deformations of standard discontinuous groups for pseudo-Riemannian homogeneous spaces
|
106 pages (minor corrections)
| null | null |
RIKEN-iTHEMS-Report-25
|
math.DG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Let $X=G/H$ be a homogeneous space of a Lie group $G$. When the isotropy subgroup $H$ is non-compact, a discrete subgroup $\Gamma$ may fail to act properly discontinuously on $X$. In this article, we address the following question: in the setting where $G$ and $H$ are reductive Lie groups and $\Gamma \backslash X$ is a standard quotient, to what extent can one deform the discrete subgroup $\Gamma$ while preserving the proper discontinuity of the action on $X$?
We provide several classification results, including conditions under which local rigidity holds for compact standard quotients $\Gamma\backslash X$, when a standard quotient can be deformed into a non-standard quotient, a characterization of the largest Zariski-closure of discontinuous groups under small deformations, and conditions under which Zariski-dense deformations occur.
|
[
{
"version": "v1",
"created": "Fri, 04 Jul 2025 11:05:21 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 14:32:45 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kannaka",
"Kazuki",
""
],
[
"Kobayashi",
"Toshiyuki",
""
]
] |
2507.03715
|
Christopher Pierce
|
Christopher J. Pierce, Tianyu Wang, Dmitri Kalinin, Andrew Zangwill, and Daniel I. Goldman
|
Aubry-Andr\'e Localization Transition for an Active Undulator
| null | null | null | null |
physics.bio-ph
|
http://creativecommons.org/licenses/by/4.0/
|
The transport of deformable self-propelling objects like bacteria, worms, snakes, and robots through heterogeneous environments is poorly understood. In this paper, we use experiment, simulation, and theory to study a snake-like robot as it undulates without sensory feedback through a narrow channel containing a linear array of boulder-like hemispherical obstacles. The profile of the boulder landscape approximates a one-dimensional potential introduced by Aubry and Andr\'e (AA) to study wave function localization in aperiodic lattices. The AA model provides a deterministically disordered alternative to the better-known phenomenon of Anderson localization, which occurs in truly random disordered lattices. When the boulder landscape is strictly periodic, the robot can pass completely through the channel. But if the landscape is sufficiently aperiodic, the robot becomes trapped and fails to exit the channel. The metrics we use to quantify this transition -- including exponential distributions of robot position when localized -- agree well with earlier experimental and theoretical work on a localization transition that occurs when quantum waves interact with the AA potential. A theoretical treatment of the robot's motion using resistive force theory modified to include spatially varying drag forces reproduces the behavior we observe. Further, our results indicate that the transition is generated by large fluctuations in the driving torques required for self-propulsion. These results point to a potentially fundamental connection between classical and quantum wave mechanics and the locomotion of undulators. Our study illustrates how analogies with models from condensed matter physics and wave optics can lead to the discovery of principles of self-propulsion in non-periodic landscapes.
|
[
{
"version": "v1",
"created": "Fri, 04 Jul 2025 17:05:10 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 19:14:38 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Pierce",
"Christopher J.",
""
],
[
"Wang",
"Tianyu",
""
],
[
"Kalinin",
"Dmitri",
""
],
[
"Zangwill",
"Andrew",
""
],
[
"Goldman",
"Daniel I.",
""
]
] |
2507.03737
|
Chong Cheng
|
Chong Cheng, Sicheng Yu, Zijian Wang, Yifan Zhou, Hao Wang
|
Outdoor Monocular SLAM with Global Scale-Consistent 3D Gaussian Pointmaps
|
Accepted by ICCV2025
| null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
3D Gaussian Splatting (3DGS) has become a popular solution in SLAM due to its high-fidelity and real-time novel view synthesis performance. However, some previous 3DGS SLAM methods employ a differentiable rendering pipeline for tracking, lack geometric priors in outdoor scenes. Other approaches introduce separate tracking modules, but they accumulate errors with significant camera movement, leading to scale drift. To address these challenges, we propose a robust RGB-only outdoor 3DGS SLAM method: S3PO-GS. Technically, we establish a self-consistent tracking module anchored in the 3DGS pointmap, which avoids cumulative scale drift and achieves more precise and robust tracking with fewer iterations. Additionally, we design a patch-based pointmap dynamic mapping module, which introduces geometric priors while avoiding scale ambiguity. This significantly enhances tracking accuracy and the quality of scene reconstruction, making it particularly suitable for complex outdoor environments. Our experiments on the Waymo, KITTI, and DL3DV datasets demonstrate that S3PO-GS achieves state-of-the-art results in novel view synthesis and outperforms other 3DGS SLAM methods in tracking accuracy. Project page: https://3dagentworld.github.io/S3PO-GS/.
|
[
{
"version": "v1",
"created": "Fri, 04 Jul 2025 17:56:43 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 16:21:11 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Cheng",
"Chong",
""
],
[
"Yu",
"Sicheng",
""
],
[
"Wang",
"Zijian",
""
],
[
"Zhou",
"Yifan",
""
],
[
"Wang",
"Hao",
""
]
] |
2507.04599
|
Yongjia Ma
|
Jiahui Yang, Yongjia Ma, Donglin Di, Hao Li, Wei Chen, Yan Xie, Jianxun Cui, Xun Yang, Wangmeng Zuo
|
QR-LoRA: Efficient and Disentangled Fine-tuning via QR Decomposition for Customized Generation
|
ICCV 2025, 30 pages, 26 figures
| null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Existing text-to-image models often rely on parameter fine-tuning techniques such as Low-Rank Adaptation (LoRA) to customize visual attributes. However, when combining multiple LoRA models for content-style fusion tasks, unstructured modifications of weight matrices often lead to undesired feature entanglement between content and style attributes. We propose QR-LoRA, a novel fine-tuning framework leveraging QR decomposition for structured parameter updates that effectively separate visual attributes. Our key insight is that the orthogonal Q matrix naturally minimizes interference between different visual features, while the upper triangular R matrix efficiently encodes attribute-specific transformations. Our approach fixes both Q and R matrices while only training an additional task-specific $\Delta R$ matrix. This structured design reduces trainable parameters to half of conventional LoRA methods and supports effective merging of multiple adaptations without cross-contamination due to the strong disentanglement properties between $\Delta R$ matrices. Experiments demonstrate that QR-LoRA achieves superior disentanglement in content-style fusion tasks, establishing a new paradigm for parameter-efficient, disentangled fine-tuning in generative models. The project page is available at: https://luna-ai-lab.github.io/QR-LoRA/.
|
[
{
"version": "v1",
"created": "Mon, 07 Jul 2025 01:31:01 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 09:12:08 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Yang",
"Jiahui",
""
],
[
"Ma",
"Yongjia",
""
],
[
"Di",
"Donglin",
""
],
[
"Li",
"Hao",
""
],
[
"Chen",
"Wei",
""
],
[
"Xie",
"Yan",
""
],
[
"Cui",
"Jianxun",
""
],
[
"Yang",
"Xun",
""
],
[
"Zuo",
"Wangmeng",
""
]
] |
2507.04600
|
Zhipeng Liu
|
Zhipeng Liu, Peibo Duan, Binwu Wang, Xuan Tang, Qi Chu, Changsheng Zhang, Yongsheng Huang, Bin Zhang
|
DisMS-TS: Eliminating Redundant Multi-Scale Features for Time Series Classification
|
This paper has been accepted for presentation at the ACM International Conference on Multimedia (ACM MM 2025)
| null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Real-world time series typically exhibit complex temporal variations, making the time series classification task notably challenging. Recent advancements have demonstrated the potential of multi-scale analysis approaches, which provide an effective solution for capturing these complex temporal patterns. However, existing multi-scale analysis-based time series prediction methods fail to eliminate redundant scale-shared features across multi-scale time series, resulting in the model over- or under-focusing on scale-shared features. To address this issue, we propose a novel end-to-end Disentangled Multi-Scale framework for Time Series classification (DisMS-TS). The core idea of DisMS-TS is to eliminate redundant shared features in multi-scale time series, thereby improving prediction performance. Specifically, we propose a temporal disentanglement module to capture scale-shared and scale-specific temporal representations, respectively. Subsequently, to effectively learn both scale-shared and scale-specific temporal representations, we introduce two regularization terms that ensure the consistency of scale-shared representations and the disparity of scale-specific representations across all temporal scales. Extensive experiments conducted on multiple datasets validate the superiority of DisMS-TS over its competitive baselines, with the accuracy improvement up to 9.71%.
|
[
{
"version": "v1",
"created": "Mon, 07 Jul 2025 01:35:55 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 09:29:08 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Liu",
"Zhipeng",
""
],
[
"Duan",
"Peibo",
""
],
[
"Wang",
"Binwu",
""
],
[
"Tang",
"Xuan",
""
],
[
"Chu",
"Qi",
""
],
[
"Zhang",
"Changsheng",
""
],
[
"Huang",
"Yongsheng",
""
],
[
"Zhang",
"Bin",
""
]
] |
2507.05109
|
Rafael Fuentes
|
J. R. Fuentes, Christopher R. Mankovich, and Ankan Sur
|
An Energy Perspective of Core Erosion in Gas Giant Planets
|
Accepted for Publication in ApJ Letters
|
ApJL 988 L49 (2025)
|
10.3847/2041-8213/adef0a
| null |
astro-ph.EP astro-ph.SR
|
http://creativecommons.org/licenses/by/4.0/
|
Juno and Cassini have shown that Jupiter and Saturn likely contain extended gradients of heavy elements. Yet, how these gradients can survive over billions of years remains an open question. Classical convection theories predict rapid mixing and homogenization, which would erase such gradients on timescales far shorter than the planets' ages. To address this, we estimate the energy required to erode both dense and fuzzy cores, and compare it to what the planet can realistically supply. If the entire cooling budget is available to drive mixing, then even a compact core can, in principle, be destroyed. But if mixing is limited to the thermal energy near the core, which is another plausible scenario, the energy falls short. In that case, Jupiter can erode a fuzzy core by up to approximately $10~\mearth$, but a compact one remains intact. Saturn's core is more robust. Even in the fuzzy case, only about $1~\mearth$ is lost, and if the core is compact, erosion is negligible. The outcome depends sensitively on the assumed initial temperature and entropy profiles. Hotter and more superadiabatic interiors are more prone to mixing. We suggest that 3D simulations of convection driven from above, with realistic stratification and enough depth (i.e., many density scale heights) would be of great interest to further constrain the energy budget for core erosion.
|
[
{
"version": "v1",
"created": "Mon, 07 Jul 2025 15:27:42 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Fuentes",
"J. R.",
""
],
[
"Mankovich",
"Christopher R.",
""
],
[
"Sur",
"Ankan",
""
]
] |
2507.05709
|
Jing Deng
|
Jing Deng, Lei Fan, Qichao Hou, Xingfang luo, Chun-Fang Rao, Yuan-Feng Zhu
|
Low-loss terahertz negative curvature suspended-core fiber
| null | null | null | null |
physics.optics
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Inspired by the design concept of negative curvature hollow-core fibers, this paper presents an innovative negative curvature suspended-core THz fiber. Compared to traditional suspended-core fibers, all structural units of this fiber are designed with circular boundaries, effectively avoiding the issues of insufficient mechanical strength and manufacturing difficulties associated with the wide and ultra-thin rectangular support arms in traditional structures. The numerical simulation using the full-vector finite element method shows that the optical fiber loss is as low as 0.0868dB/cm (0.02cm-1) in 0.4-0.9THz, and the low loss bandwidth is 0.43THz. In addition, by adjusting the structural parameters of the fiber, near zero flat dispersion of -0.15-0.05 ps/THz/cm can be achieved in the range of 0.4-0.9 THz. The fiber exhibits excellent characteristics of low loss, wide bandwidth, and low dispersion, theoretically opening a new research path for the design of low loss THz fibers. This will provide important component support for the application and development of THz technology in fields such as communication, sensing, and imaging.
|
[
{
"version": "v1",
"created": "Tue, 08 Jul 2025 06:40:14 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 06:44:42 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Deng",
"Jing",
""
],
[
"Fan",
"Lei",
""
],
[
"Hou",
"Qichao",
""
],
[
"luo",
"Xingfang",
""
],
[
"Rao",
"Chun-Fang",
""
],
[
"Zhu",
"Yuan-Feng",
""
]
] |
2507.06024
|
Sofya Maslovskaya
|
Michael Konopik, Sigrid Leyendecker, Sofya Maslovskaya, Sina Ober-Bl\"obaum, Rodrigo T. Sato Mart\'in de Almagro
|
Second order optimality conditions in a new Lagrangian formulation for optimal control problems
| null | null | null | null |
math.OC
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
It has been shown recently that optimal control problems with the dynamical constraint given by a second order system admit a regular Lagrangian formulation. This implies that the optimality conditions can be obtained in a new form based on the variational approach. In this paper we extend the first order necessary optimality conditions obtained previously to second order optimality conditions. This results in a complete characterization of the optimality conditions in a new Lagrangian form.
|
[
{
"version": "v1",
"created": "Tue, 08 Jul 2025 14:31:16 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 13:23:43 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Konopik",
"Michael",
""
],
[
"Leyendecker",
"Sigrid",
""
],
[
"Maslovskaya",
"Sofya",
""
],
[
"Ober-Blöbaum",
"Sina",
""
],
[
"de Almagro",
"Rodrigo T. Sato Martín",
""
]
] |
2507.06174
|
Koki Yamane
|
Koki Yamane, Yunhan Li, Masashi Konosu, Koki Inami, Junji Oaki, Sho Sakaino, Toshiaki Tsuji
|
Fast Bilateral Teleoperation and Imitation Learning Using Sensorless Force Control via Accurate Dynamics Model
|
20 pages, 9 figures, Submitted to CoRL 2025
| null | null | null |
cs.RO cs.AI cs.SY eess.SY
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In recent years, the advancement of imitation learning has led to increased interest in teleoperating low-cost manipulators to collect demonstration data. However, most existing systems rely on unilateral control, which only transmits target position values. While this approach is easy to implement and suitable for slow, non-contact tasks, it struggles with fast or contact-rich operations due to the absence of force feedback. This work demonstrates that fast teleoperation with force feedback is feasible even with force-sensorless, low-cost manipulators by leveraging 4-channel bilateral control. Based on accurately identified manipulator dynamics, our method integrates nonlinear terms compensation, velocity and external force estimation, and variable gain corresponding to inertial variation. Furthermore, using data collected by 4-channel bilateral control, we show that incorporating force information into both the input and output of learned policies improves performance in imitation learning. These results highlight the practical effectiveness of our system for high-fidelity teleoperation and data collection on affordable hardware.
|
[
{
"version": "v1",
"created": "Tue, 08 Jul 2025 16:54:34 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Jul 2025 16:53:31 GMT"
},
{
"version": "v3",
"created": "Thu, 17 Jul 2025 16:43:46 GMT"
},
{
"version": "v4",
"created": "Sat, 19 Jul 2025 00:20:11 GMT"
},
{
"version": "v5",
"created": "Thu, 24 Jul 2025 00:40:26 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Yamane",
"Koki",
""
],
[
"Li",
"Yunhan",
""
],
[
"Konosu",
"Masashi",
""
],
[
"Inami",
"Koki",
""
],
[
"Oaki",
"Junji",
""
],
[
"Sakaino",
"Sho",
""
],
[
"Tsuji",
"Toshiaki",
""
]
] |
2507.06646
|
Zicong Peng
|
Zicong Peng, Yicheng Zhan, Josef Spjut, Kaan Ak\c{s}it
|
Assessing Learned Models for Phase-only Hologram Compression
|
SIGGRAPH 2025 Poster
| null |
10.1145/3721250.3742993
| null |
cs.GR
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
We evaluate the performance of four common learned models utilizing INR and VAE structures for compressing phase-only holograms in holographic displays. The evaluated models include a vanilla MLP, SIREN, and FilmSIREN, with TAESD as the representative VAE model. Our experiments reveal that a pretrained image VAE, TAESD, with 2.2M parameters struggles with phase-only hologram compression, revealing the need for task-specific adaptations. Among the INRs, SIREN with 4.9k parameters achieves %40 compression with high quality in the reconstructed 3D images (PSNR = 34.54 dB). These results emphasize the effectiveness of INRs and identify the limitations of pretrained image compression VAEs for hologram compression task.
|
[
{
"version": "v1",
"created": "Wed, 09 Jul 2025 08:19:44 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 13:01:37 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Peng",
"Zicong",
""
],
[
"Zhan",
"Yicheng",
""
],
[
"Spjut",
"Josef",
""
],
[
"Akşit",
"Kaan",
""
]
] |
2507.06735
|
Xue Wang
|
Guan Zheng, Xue Wang, Wenhua Qian, Peng Liu, Runzhuo Ma
|
Residual Prior-driven Frequency-aware Network for Image Fusion
| null | null |
10.1145/3746027.3754944
| null |
cs.CV cs.LG cs.MM
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Image fusion aims to integrate complementary information across modalities to generate high-quality fused images, thereby enhancing the performance of high-level vision tasks. While global spatial modeling mechanisms show promising results, constructing long-range feature dependencies in the spatial domain incurs substantial computational costs. Additionally, the absence of ground-truth exacerbates the difficulty of capturing complementary features effectively. To tackle these challenges, we propose a Residual Prior-driven Frequency-aware Network, termed as RPFNet. Specifically, RPFNet employs a dual-branch feature extraction framework: the Residual Prior Module (RPM) extracts modality-specific difference information from residual maps, thereby providing complementary priors for fusion; the Frequency Domain Fusion Module (FDFM) achieves efficient global feature modeling and integration through frequency-domain convolution. Additionally, the Cross Promotion Module (CPM) enhances the synergistic perception of local details and global structures through bidirectional feature interaction. During training, we incorporate an auxiliary decoder and saliency structure loss to strengthen the model's sensitivity to modality-specific differences. Furthermore, a combination of adaptive weight-based frequency contrastive loss and SSIM loss effectively constrains the solution space, facilitating the joint capture of local details and global features while ensuring the retention of complementary information. Extensive experiments validate the fusion performance of RPFNet, which effectively integrates discriminative features, enhances texture details and salient objects, and can effectively facilitate the deployment of the high-level vision task.
|
[
{
"version": "v1",
"created": "Wed, 09 Jul 2025 10:48:00 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 13:57:08 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zheng",
"Guan",
""
],
[
"Wang",
"Xue",
""
],
[
"Qian",
"Wenhua",
""
],
[
"Liu",
"Peng",
""
],
[
"Ma",
"Runzhuo",
""
]
] |
2507.07464
|
Chang-Hwan Son
|
Chang-Hwan Son
|
Degradation-Agnostic Statistical Facial Feature Transformation for Blind Face Restoration in Adverse Weather Conditions
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
With the increasing deployment of intelligent CCTV systems in outdoor environments, there is a growing demand for face recognition systems optimized for challenging weather conditions. Adverse weather significantly degrades image quality, which in turn reduces recognition accuracy. Although recent face image restoration (FIR) models based on generative adversarial networks (GANs) and diffusion models have shown progress, their performance remains limited due to the lack of dedicated modules that explicitly address weather-induced degradations. This leads to distorted facial textures and structures. To address these limitations, we propose a novel GAN-based blind FIR framework that integrates two key components: local Statistical Facial Feature Transformation (SFFT) and Degradation-Agnostic Feature Embedding (DAFE). The local SFFT module enhances facial structure and color fidelity by aligning the local statistical distributions of low-quality (LQ) facial regions with those of high-quality (HQ) counterparts. Complementarily, the DAFE module enables robust statistical facial feature extraction under adverse weather conditions by aligning LQ and HQ encoder representations, thereby making the restoration process adaptive to severe weather-induced degradations. Experimental results demonstrate that the proposed degradation-agnostic SFFT model outperforms existing state-of-the-art FIR methods based on GAN and diffusion models, particularly in suppressing texture distortions and accurately reconstructing facial structures. Furthermore, both the SFFT and DAFE modules are empirically validated in enhancing structural fidelity and perceptual quality in face restoration under challenging weather scenarios.
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 06:31:26 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 12:13:12 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Son",
"Chang-Hwan",
""
]
] |
2507.07620
|
Marc Lafon
|
Marc Lafon, Yannis Karmim, Julio Silva-Rodr\'iguez, Paul Couairon, Cl\'ement Rambour, Rapha\"el Fournier-Sniehotta, Ismail Ben Ayed, Jose Dolz, Nicolas Thome
|
ViLU: Learning Vision-Language Uncertainties for Failure Prediction
| null |
International Conference on Computer Vision, ICCV 2025
| null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Reliable Uncertainty Quantification (UQ) and failure prediction remain open challenges for Vision-Language Models (VLMs). We introduce ViLU, a new Vision-Language Uncertainty quantification framework that contextualizes uncertainty estimates by leveraging all task-relevant textual representations. ViLU constructs an uncertainty-aware multi-modal representation by integrating the visual embedding, the predicted textual embedding, and an image-conditioned textual representation via cross-attention. Unlike traditional UQ methods based on loss prediction, ViLU trains an uncertainty predictor as a binary classifier to distinguish correct from incorrect predictions using a weighted binary cross-entropy loss, making it loss-agnostic. In particular, our proposed approach is well-suited for post-hoc settings, where only vision and text embeddings are available without direct access to the model itself. Extensive experiments on diverse datasets show the significant gains of our method compared to state-of-the-art failure prediction methods. We apply our method to standard classification datasets, such as ImageNet-1k, as well as large-scale image-caption datasets like CC12M and LAION-400M. Ablation studies highlight the critical role of our architecture and training in achieving effective uncertainty quantification. Our code is publicly available and can be found here: https://github.com/ykrmm/ViLU.
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 10:41:13 GMT"
},
{
"version": "v2",
"created": "Fri, 11 Jul 2025 15:41:56 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 09:03:07 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Lafon",
"Marc",
""
],
[
"Karmim",
"Yannis",
""
],
[
"Silva-Rodríguez",
"Julio",
""
],
[
"Couairon",
"Paul",
""
],
[
"Rambour",
"Clément",
""
],
[
"Fournier-Sniehotta",
"Raphaël",
""
],
[
"Ayed",
"Ismail Ben",
""
],
[
"Dolz",
"Jose",
""
],
[
"Thome",
"Nicolas",
""
]
] |
2507.07665
|
Chen Yuan
|
Chen Yuan, Zhen Zhong, and Qing-Guo Huang
|
Whispers from the Early Universe: The Ringdown of Primordial Black Holes
|
6 pages, 3 figures
| null | null | null |
astro-ph.CO gr-qc hep-th
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We investigate the stochastic gravitational wave background (SGWB) generated by the ringdown phase of primordial black holes (PBHs) formed in the early universe. As the ringdown signal is independent of the PBH formation mechanism, the resulting SGWB offers a model-independent probe of PBHs. We numerically compute the ringdown waveform and derive the corresponding SGWB. We show that such a signal could be detected by future pulsar timing arrays (PTAs) for PBHs heavier than the solar mass. Additionally, we evaluate the SGWB from binary PBH mergers and demonstrate that it lies within the sensitivity bands of next-generation ground-based interferometers such as Cosmic Explorer and Einstein Telescope, suggesting a multi-band observational strategy for probing the PBH dark matter scenario.
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 11:46:29 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 18:01:13 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Yuan",
"Chen",
""
],
[
"Zhong",
"Zhen",
""
],
[
"Huang",
"Qing-Guo",
""
]
] |
2507.07769
|
Ruohong Liu
|
Ruohong Liu, Jack Umenberger, Yize Chen
|
BEAVER: Building Environments with Assessable Variation for Evaluating Multi-Objective Reinforcement Learning
|
Accepted at the Workshop on Computational Optimization of Buildings (ICML CO-BUILD), 42nd International Conference on Machine Learning (ICML 2025), Vancouver, Canada
| null | null | null |
cs.LG cs.SY eess.SY
|
http://creativecommons.org/licenses/by/4.0/
|
Recent years have seen significant advancements in designing reinforcement learning (RL)-based agents for building energy management. While individual success is observed in simulated or controlled environments, the scalability of RL approaches in terms of efficiency and generalization across building dynamics and operational scenarios remains an open question. In this work, we formally characterize the generalization space for the cross-environment, multi-objective building energy management task, and formulate the multi-objective contextual RL problem. Such a formulation helps understand the challenges of transferring learned policies across varied operational contexts such as climate and heat convection dynamics under multiple control objectives such as comfort level and energy consumption. We provide a principled way to parameterize such contextual information in realistic building RL environments, and construct a novel benchmark to facilitate the evaluation of generalizable RL algorithms in practical building control tasks. Our results show that existing multi-objective RL methods are capable of achieving reasonable trade-offs between conflicting objectives. However, their performance degrades under certain environment variations, underscoring the importance of incorporating dynamics-dependent contextual information into the policy learning process.
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 13:54:38 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 10:44:28 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Liu",
"Ruohong",
""
],
[
"Umenberger",
"Jack",
""
],
[
"Chen",
"Yize",
""
]
] |
2507.07876
|
Timur Aslyamov Dr
|
Timur Aslyamov and Massimiliano Esposito
|
Excess Observables Reveal Nonreciprocity in Integrated Covariance
| null | null | null | null |
cond-mat.stat-mech
|
http://creativecommons.org/licenses/by/4.0/
|
Near equilibrium, the symmetric part of the time-integrated steady-state covariance, i.e., the time integral of correlation functions, is governed by the fluctuation-dissipation theorem, while the antisymmetric part vanishes due to Onsager reciprocity. Far from equilibrium, where these principles no longer apply, we develop a unified formalism for both symmetric and antisymmetric components of integrated covariances. We derive exact, computationally tractable expressions for these quantities, valid in arbitrary nonequilibrium steady states of Markov jump processes. Both components are expressed in terms of excess observables, a notion central to both statistical physics and reinforcement learning. Furthermore, we establish thermodynamic upper bounds for these covariances in terms of entropy production, dynamical activity, and cycle affinities.
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 15:59:44 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:05:53 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Aslyamov",
"Timur",
""
],
[
"Esposito",
"Massimiliano",
""
]
] |
2507.07893
|
Mingda Zhang
|
Mingda Zhang, Na Zhao, Jianglong Qing, Qing xu, Kaiwen Pan, Ting luo
|
An Integrated Framework of Prompt Engineering and Multidimensional Knowledge Graphs for Legal Dispute Analysis
|
19 pages,3 figures
| null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Legal dispute analysis is crucial for intelligent legal assistance systems. However, current LLMs face significant challenges in understanding complex legal concepts, maintaining reasoning consistency, and accurately citing legal sources. This research presents a framework combining prompt engineering with multidimensional knowledge graphs to improve LLMs' legal dispute analysis. Specifically, the framework includes a three-stage hierarchical prompt structure (task definition, knowledge background, reasoning guidance) along with a three-layer knowledge graph (legal ontology, representation, instance layers). Additionally, four supporting methods enable precise legal concept retrieval: direct code matching, semantic vector similarity, ontology path reasoning, and lexical segmentation. Through extensive testing, results show major improvements: sensitivity increased by 9.9%-13.8%, specificity by 4.8%-6.7%, and citation accuracy by 22.4%-39.7%. As a result, the framework provides better legal analysis and understanding of judicial logic, thus offering a new technical method for intelligent legal assistance systems.
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 16:22:41 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Jul 2025 14:41:16 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 13:52:51 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zhang",
"Mingda",
""
],
[
"Zhao",
"Na",
""
],
[
"Qing",
"Jianglong",
""
],
[
"xu",
"Qing",
""
],
[
"Pan",
"Kaiwen",
""
],
[
"luo",
"Ting",
""
]
] |
2507.07966
|
Chen Yukang
|
Yukang Chen, Wei Huang, Baifeng Shi, Qinghao Hu, Hanrong Ye, Ligeng Zhu, Zhijian Liu, Pavlo Molchanov, Jan Kautz, Xiaojuan Qi, Sifei Liu, Hongxu Yin, Yao Lu, Song Han
|
Scaling RL to Long Videos
|
Code at https://github.com/NVlabs/Long-RL and model at https://huggingface.co/Efficient-Large-Model/LongVILA-R1-7B
| null | null | null |
cs.CV cs.AI cs.CL
|
http://creativecommons.org/licenses/by-sa/4.0/
|
We introduce a full-stack framework that scales up reasoning in vision-language models (VLMs) to long videos, leveraging reinforcement learning. We address the unique challenges of long video reasoning by integrating three critical components: (1) a large-scale dataset, LongVideo-Reason, comprising 104K long video QA pairs with high-quality reasoning annotations across diverse domains such as sports, games, and vlogs; (2) a two-stage training pipeline that extends VLMs with chain-of-thought supervised fine-tuning (CoT-SFT) and reinforcement learning (RL); and (3) a training infrastructure for long video RL, named Multi-modal Reinforcement Sequence Parallelism (MR-SP), which incorporates sequence parallelism and a vLLM-based engine tailored for long video, using cached video embeddings for efficient rollout and prefilling. In our experiments, LongVILA-R1-7B achieves strong performance on video benchmarks, reaching 65.0% and 70.7% accuracy on VideoMME without and with subtitles, respectively, and consistently outperforming LongVILA-R1 across multiple benchmarks. Moreover, LongVILA-R1 shows steady performance improvements as the number of input video frames increases. Notably, our MR-SP system achieves up to 2.1x speedup on long video RL training. In addition, we release our training system for public availability that supports RL training on various modalities (video, text, and audio), various models (VILA and Qwen series), and even image and video generation models. On a single A100 node (8 GPUs), it supports RL training on hour-long videos (e.g., 3,600 frames / around 256k tokens).
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 17:47:40 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 17:20:41 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Chen",
"Yukang",
""
],
[
"Huang",
"Wei",
""
],
[
"Shi",
"Baifeng",
""
],
[
"Hu",
"Qinghao",
""
],
[
"Ye",
"Hanrong",
""
],
[
"Zhu",
"Ligeng",
""
],
[
"Liu",
"Zhijian",
""
],
[
"Molchanov",
"Pavlo",
""
],
[
"Kautz",
"Jan",
""
],
[
"Qi",
"Xiaojuan",
""
],
[
"Liu",
"Sifei",
""
],
[
"Yin",
"Hongxu",
""
],
[
"Lu",
"Yao",
""
],
[
"Han",
"Song",
""
]
] |
2507.08017
|
Pierre Beckmann
|
Pierre Beckmann and Matthieu Queloz
|
Mechanistic Indicators of Understanding in Large Language Models
|
32 pages
| null | null | null |
cs.CL cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Recent findings in mechanistic interpretability (MI), the field probing the inner workings of Large Language Models (LLMs), challenge the view that these models rely solely on superficial statistics. We offer an accessible synthesis of these findings that doubles as an introduction to MI while integrating these findings within a novel theoretical framework for thinking about machine understanding. We argue that LLMs develop internal structures that are functionally analogous to the kind of understanding that consists in seeing connections. To sharpen this idea, we propose a three-tiered conception of understanding. First, conceptual understanding emerges when a model forms "features" as directions in latent space, learning the connections between diverse manifestations of something. Second, state-of-the-world understanding emerges when a model learns contingent factual connections between features and dynamically tracks changes in the world. Third, principled understanding emerges when a model ceases to rely on a collection of memorized facts and discovers a "circuit" connecting these facts. However, these forms of understanding remain radically different from human understanding, as the phenomenon of "parallel mechanisms" shows. We conclude that the debate should move beyond the yes-or-no question of whether LLMs understand to investigate how their strange minds work and forge conceptions that fit them.
|
[
{
"version": "v1",
"created": "Mon, 07 Jul 2025 20:26:31 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Jul 2025 11:46:41 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 12:23:53 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Beckmann",
"Pierre",
""
],
[
"Queloz",
"Matthieu",
""
]
] |
2507.08351
|
Peter Schmelcher
|
Peter Schmelcher
|
Computational Study of the Spectral Behaviour of Different Isospectrally Patterned Lattices
| null | null | null | null |
quant-ph cond-mat.other
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We perform a computational spectral analysis of different isospectrally patterned lattices (IPL). Having been introduced very recently, the lattice Hamiltonian of IPL consist of coupled cells which possess all the same set of eigenvalues. The latter is achieved in a controllable manner by parametrizing the cells via the phases of the involved orthogonal (or unitary) transformations. This opens the doorway of systematically designing lattice Hamiltonians with unique properties by choosing correspondingly varying phases across the lattice. Here we focus on two-dimensional cells and explore symmetric as well as asymmetric IPL w.r.t. a given center. A tunable fraction of localized vs. delocalized eigenstates belonging to the three subdomains of the corresponding energy bands is demonstrated and analyzed with different measures of localization. In the asymmetric case the center of localization can be shifted arbitrarily by shifting the underlying phase grid. Introducing a complete phase revolution leads for low and high energies to two well-separated branches of localized states which finally merge with increasing energy into the branch of delocalized states. Remarkably, the localized states appear in near-degenerate pairs and this near-degeneracy is lifted upon entering the delocalization regime. A corresponding generalization to several phase revolutions is provided showing a characteristic nodal pattern among the near-degenerate eigenstates.
|
[
{
"version": "v1",
"created": "Fri, 11 Jul 2025 06:56:50 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 07:44:27 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Schmelcher",
"Peter",
""
]
] |
2507.08449
|
Yakun Wang
|
Yakun Wang and Yuan Liu and Bin Wu
|
The stability of bi-polarization on dynamical directed graphs: an emergent game perspective
| null | null | null | null |
nlin.AO physics.soc-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper proposes a co-evolutionary model of directed graphs and three opinions, i.e., conservative$(+)$, neutral$(\odot)$ and liberal$(-)$. Agents update both opinions and social relationships with bias. We find that an emergent game suffices to predict the stability of bi-polarization under a rare opinion updating limit and a large system size limit. The bi-polarization is stable if and only if the emergent game has an internal Nash equilibrium. The necessary and sufficient condition is explained by both risk dominance and evolutionary stability. This game approach facilitates us to reveal the stability of bi-polarization in empirical systems. Our work fosters the understanding of opinion formation for controversial topics, and shows a deep connection between opinion dynamics and evolutionary game theory.
|
[
{
"version": "v1",
"created": "Fri, 11 Jul 2025 09:43:41 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Wang",
"Yakun",
""
],
[
"Liu",
"Yuan",
""
],
[
"Wu",
"Bin",
""
]
] |
2507.08513
|
Liu He
|
Liu He, Xiao Zeng, Yizhi Song, Albert Y. C. Chen, Lu Xia, Shashwat Verma, Sankalp Dayal, Min Sun, Cheng-Hao Kuo, Daniel Aliaga
|
Advancing Multimodal LLMs by Large-Scale 3D Visual Instruction Dataset Generation
| null | null | null | null |
cs.GR cs.CV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Multimodal Large Language Models (MLLMs) struggle with accurately capturing camera-object relations, especially for object orientation, camera viewpoint, and camera shots. This stems from the fact that existing MLLMs are trained on images with limited diverse camera-object relations and corresponding textual descriptions. To address this, we propose a synthetic generation pipeline to create large-scale 3D visual instruction datasets. Our framework takes 3D assets as input and uses rendering and diffusion-based image generation models to create photorealistic images preserving precise camera-object relations. Additionally, large language models (LLMs) are used to generate text prompts for guiding visual instruction tuning and controlling image generation. We create Ultimate3D, a dataset of 240K VQAs with precise camera-object annotations, and corresponding benchmark. MLLMs fine-tuned on our proposed dataset outperform commercial models by a large margin, achieving an average accuracy improvement of 33.4% on camera-object relation recognition tasks. Our code, dataset, and benchmark will contribute to broad MLLM applications.
|
[
{
"version": "v1",
"created": "Fri, 11 Jul 2025 12:00:10 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 22:34:55 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"He",
"Liu",
""
],
[
"Zeng",
"Xiao",
""
],
[
"Song",
"Yizhi",
""
],
[
"Chen",
"Albert Y. C.",
""
],
[
"Xia",
"Lu",
""
],
[
"Verma",
"Shashwat",
""
],
[
"Dayal",
"Sankalp",
""
],
[
"Sun",
"Min",
""
],
[
"Kuo",
"Cheng-Hao",
""
],
[
"Aliaga",
"Daniel",
""
]
] |
2507.08621
|
Marcin Pietron
|
Marcin Pietro\'n, Rafa{\l} Olszowski, Jakub Gomu{\l}ka, Filip Gampel, Andrzej Tomski
|
A comprehensive study of LLM-based argument classification: from LLAMA through GPT-4o to Deepseek-R1
| null | null | null | null |
cs.CL cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Argument mining (AM) is an interdisciplinary research field that integrates insights from logic, philosophy, linguistics, rhetoric, law, psychology, and computer science. It involves the automatic identification and extraction of argumentative components, such as premises and claims, and the detection of relationships between them, such as support, attack, or neutrality. Recently, the field has advanced significantly, especially with the advent of large language models (LLMs), which have enhanced the efficiency of analyzing and extracting argument semantics compared to traditional methods and other deep learning models. There are many benchmarks for testing and verifying the quality of LLM, but there is still a lack of research and results on the operation of these models in publicly available argument classification databases. This paper presents a study of a selection of LLM's, using diverse datasets such as Args.me and UKP. The models tested include versions of GPT, Llama, and DeepSeek, along with reasoning-enhanced variants incorporating the Chain-of-Thoughts algorithm. The results indicate that ChatGPT-4o outperforms the others in the argument classification benchmarks. In case of models incorporated with reasoning capabilities, the Deepseek-R1 shows its superiority. However, despite their superiority, GPT-4o and Deepseek-R1 still make errors. The most common errors are discussed for all models. To our knowledge, the presented work is the first broader analysis of the mentioned datasets using LLM and prompt algorithms. The work also shows some weaknesses of known prompt algorithms in argument analysis, while indicating directions for their improvement. The added value of the work is the in-depth analysis of the available argument datasets and the demonstration of their shortcomings.
|
[
{
"version": "v1",
"created": "Fri, 11 Jul 2025 14:23:40 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 11:49:06 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Pietroń",
"Marcin",
""
],
[
"Olszowski",
"Rafał",
""
],
[
"Gomułka",
"Jakub",
""
],
[
"Gampel",
"Filip",
""
],
[
"Tomski",
"Andrzej",
""
]
] |
2507.08723
|
Bel\'en Andrada
|
Bel\'en Andrada (for the Pierre Auger Collaboration)
|
Monitoring and performance of AugerPrime
|
Presented at the 39th International Cosmic Ray Conference (ICRC 2025). 12 pages, 6 figures
| null | null |
PoS-ICRC2025-176
|
astro-ph.IM hep-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The upgrade of Pierre Auger Observatory, AugerPrime, is a multi-hybrid system designed to improve the sensitivity and precision of ultra-high-energy cosmic ray measurements. It includes scintillator detectors positioned both atop the enhanced Water-Cherenkov detectors and buried nearby for direct muon measurements, along with radio and fluorescence detectors. In this contribution, we present an overview of the monitoring tools developed for all the components of AugerPrime, focusing on real-time performance assessment and long-term stability metrics. By continuously tracking key parameters, we can identify potential issues early, enabling timely interventions and improving overall data quality. These strategies are crucial for maintaining the long-term reliability of the measurements taken at the Auger Observatory and providing high-quality data for cosmic ray research in the coming decades.
|
[
{
"version": "v1",
"created": "Fri, 11 Jul 2025 16:22:51 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Andrada",
"Belén",
"",
"for the Pierre Auger Collaboration"
]
] |
2507.08780
|
Martin Bishop
|
Martin Bishop
|
Brauer groups of tame stacky curves and their $\mu_r$-gerbes
|
24 pages. Rewrote intro, reorganized background, removed extraneous result, mild improvements throughout
| null | null | null |
math.AG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We fit the Brauer group of a $\mu_r$-gerbe over a (possibly arbitrarily singular) stacky curve into an exact sequence and give characterizations for when it is short exact and conditions for when it splits. We also give a precise formula for the Brauer group of a $\mu_r$-gerbe in the case that the base curve is smooth.
|
[
{
"version": "v1",
"created": "Fri, 11 Jul 2025 17:42:10 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 17:30:40 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Bishop",
"Martin",
""
]
] |
2507.08893
|
Cook Hyun Kim
|
Cook Hyun Kim and B. Kahng
|
Spectral-Space RG Theory Based on Universal Scaling Relations
| null | null | null | null |
physics.soc-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Scale-free networks -- from the Internet to biological systems -- exhibit hierarchical organization that resists conventional renormalization group (RG) analysis. Their combination of scale invariance and small-world connectivity challenges standard RG methods, which rely on well-defined length scales. We resolve this challenge by formulating a spectral-space RG framework that captures both structural and dynamical scaling in complex networks. Leveraging the Laplacian eigenspectrum, we implement coarse-graining transformations unconstrained by geometry. This yields universal scaling relations connecting fractal dimensions, spectral dimensions, and degree exponents, establishing the first systematic foundation for network renormalization. A novel meta-graph reconstruction algorithm enables direct extraction of renormalized topologies from spectral data. We validate our predictions across diverse real-world networks and uncover new phenomena: evolving networks display multi-scaling behavior indicative of structural transitions, and spectral non-recursiveness reveals hidden dynamical correlations invisible in static topology. Applied to the European power grid, our method identifies latent connections between distant regions, consistent with observed fault propagation. Our results position spectral-space renormalization as a unified framework for analyzing scale-invariant networks, with broad implications for network science, infrastructure resilience, and statistical physics.
|
[
{
"version": "v1",
"created": "Thu, 10 Jul 2025 22:59:17 GMT"
},
{
"version": "v2",
"created": "Tue, 15 Jul 2025 11:46:11 GMT"
},
{
"version": "v3",
"created": "Tue, 22 Jul 2025 08:46:14 GMT"
},
{
"version": "v4",
"created": "Thu, 24 Jul 2025 00:12:21 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kim",
"Cook Hyun",
""
],
[
"Kahng",
"B.",
""
]
] |
2507.09006
|
Marina Scornavacche Mrs
|
Marina Scornavacche (for the Pierre Auger Collaboration)
|
Muon Signal Charge in the Underground Muon Detector of AugerPrime
|
Presented at the 39th International Cosmic Ray Conference (ICRC 2025). 12 pages, 5 figures
| null | null |
PoS-ICRC2025-389
|
physics.ins-det astro-ph.IM hep-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The Underground Muon Detector of the Pierre Auger Observatory features a calorimetric detection mode that estimates the number of muons based on signal charge measurements. This contribution provides an overview of the calibration procedure, revisiting the previously published strategy and identifying a bias introduced by the triggers used to estimate the mean charge deposited by a vertical muon. We demonstrate that calibrating underground detectors requires careful consideration of the interactions of penetrating particles through matter. In devices based on plastic scintillators, energy deposition, and thus the recorded charge, is significantly affected by knock-on electron production in the surrounding ground as muons traverse the medium. To mitigate this effect, we propose a new calibration strategy that ensures an unbiased muon estimator. This approach is applied to data collected from 2019 to 2024.
|
[
{
"version": "v1",
"created": "Fri, 11 Jul 2025 20:16:33 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Scornavacche",
"Marina",
"",
"for the Pierre Auger Collaboration"
]
] |
2507.09186
|
Minhaj Uddin Ahmad
|
Minhaj Uddin Ahmad, Akid Abrar, Sagar Dasgupta, Mizanur Rahman
|
OpenCAMS: An Open-Source Connected and Automated Mobility Co-Simulation Platform for Advancing Next-Generation Intelligent Transportation Systems Research
| null | null | null | null |
cs.SE
|
http://creativecommons.org/licenses/by/4.0/
|
We introduce OpenCAMS (Open-Source Connected and Automated Mobility Co-Simulation Platform), an open-source, synchronized, and extensible co-simulation framework that tightly couples three best-in-class simulation tools: (i) SUMO, (ii) CARLA, and (iii) OMNeT++. OpenCAMS is designed to support advanced research in transportation safety, mobility, and cybersecurity by combining the strengths of each simulation domain. Specifically, SUMO provides large-scale, microscopic traffic modeling; CARLA offers high-fidelity 3D perception, vehicle dynamics, and control simulation; and OMNeT++ enables modular, event-driven network communication, such as cellular vehicle-to-everything (C-V2X). OpenCAMS employs a time-synchronized, bidirectional coupling architecture that ensures coherent simulation progression across traffic, perception, and communication domains while preserving modularity and reproducibility. For example, CARLA can simulate and render a subset of vehicles that require detailed sensor emulation and control logic; SUMO orchestrates network-wide traffic flow, vehicle routing, and traffic signal management; and OMNeT++ dynamically maps communication nodes to both mobile entities (e.g., vehicles) and static entities (e.g., roadside units) to enable C-V2X communication. While these three simulators form the foundational core of OpenCAMS, the platform is designed to be expandable and future-proof, allowing additional simulators to be integrated on top of this core without requiring fundamental changes to the system architecture. The OpenCAMS platform is fully open-source and publicly available through its GitHub repository https://github.com/minhaj6/carla-sumo-omnetpp-cosim, providing the research community with an accessible, flexible, and collaborative environment for advancing next-generation intelligent transportation systems.
|
[
{
"version": "v1",
"created": "Sat, 12 Jul 2025 08:10:37 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Jul 2025 18:38:34 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 17:21:57 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Ahmad",
"Minhaj Uddin",
""
],
[
"Abrar",
"Akid",
""
],
[
"Dasgupta",
"Sagar",
""
],
[
"Rahman",
"Mizanur",
""
]
] |
2507.09228
|
Sinah Legner
|
Sinah Legner, Will Handley, Will Barker
|
Alleviating the Hubble tension with Torsion Condensation (TorC)
|
21 pages (main text: 12 pages), 9 figures, 7 tables, comments welcome!
| null | null | null |
astro-ph.CO gr-qc hep-th
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Constraints on the cosmological parameters of Torsion Condensation (TorC) are investigated using Planck 2018 Cosmic Microwave Background data. TorC is a case of Poincar\'e gauge theory -- a formulation of gravity motivated by the gauge field theories underlying fundamental forces in the standard model of particle physics. Unlike general relativity, TorC incorporates intrinsic torsion degrees of freedom while maintaining second-order field equations. At specific parameter values, it reduces to the $\Lambda$CDM model, providing a natural extension to standard cosmology. The base model of TorC introduces two parameters beyond those in $\Lambda$CDM: the initial value of the torsion scalar field and its time derivative -- one can absorb the latter by allowing the dark energy density to float. To constrain these parameters, `PolyChord` nested sampling algorithm is employed, interfaced via `Cobaya` with a modified version of `CAMB`. Our results indicate that TorC allows for a larger inferred Hubble constant, offering a potential resolution to the Hubble tension. Tension analysis using the $R$-statistic shows that TorC alleviates the statistical tension between the Planck 2018 and SH0Es 2020 datasets, though this improvement is not sufficient to decisively favour TorC over $\Lambda$CDM in a Bayesian model comparison. This study highlights TorC as a compelling theory of gravity, demonstrating its potential to address cosmological tensions and motivating further investigations of extended theories of gravity within a cosmological context. As current and upcoming surveys -- including Euclid, Roman Space Telescope, Vera C. Rubin Observatory, LISA, and Simons Observatory -- deliver data on gravity across all scales, they will offer critical tests of gravity models like TorC, making the present a pivotal moment for exploring extended theories of gravity.
|
[
{
"version": "v1",
"created": "Sat, 12 Jul 2025 09:56:02 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 16:50:01 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Legner",
"Sinah",
""
],
[
"Handley",
"Will",
""
],
[
"Barker",
"Will",
""
]
] |
2507.09235
|
Matija Kocbek
|
Matija Kocbek
|
Explicit geometric construction of triangle-free Ramsey graphs
|
6 pages; changed the structure of the article, added a new result
| null | null | null |
math.CO
|
http://creativecommons.org/licenses/by/4.0/
|
We describe an explicit geometric construction of a vast family of graphs without $m$-cliques with bounded independence number generalizing triangle-free Ramsey graphs described by Codenotti, Pudl\'ak and Resta and provide a new combinatorial proof for the upper bound on the independence number of the latter. We focus on triangle-free graphs and describe some families of such graphs with $n$ vertices and independence number $O(n^{\frac{2}{3}})$ which give us a constructive asymptotic lower bound $\Omega(t^{\frac{3}{2}})$ for Ramsey numbers $R(3, t)$ which achieves the best-known constructive lower bound. We describe an additional family of graphs that don't match the best-known bound but still have a polynomial independence number with regards to the number of vertices and are based on Euclidean geometry. We also present a linear $\frac{1}{2}$-approximation algorithm for finding the largest independent set that works for a significant subset of our family of graphs.
|
[
{
"version": "v1",
"created": "Sat, 12 Jul 2025 10:03:43 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:46:30 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kocbek",
"Matija",
""
]
] |
2507.09305
|
Danny Xu
|
Zhiwei Xu
|
DAA*: Deep Angular A Star for Image-based Path Planning
|
International Conference on Computer Vision (ICCV), 2025
| null | null | null |
cs.CV cs.LG eess.IV
|
http://creativecommons.org/licenses/by/4.0/
|
Path smoothness is often overlooked in path imitation learning from expert demonstrations. In this paper, we introduce a novel learning method, termed deep angular A* (DAA*), by incorporating the proposed path angular freedom (PAF) into A* to improve path similarity through adaptive path smoothness. The PAF aims to explore the effect of move angles on path node expansion by finding the trade-off between their minimum and maximum values, allowing for high adaptiveness for imitation learning. DAA* improves path optimality by closely aligning with the reference path through joint optimization of path shortening and smoothing, which correspond to heuristic distance and PAF, respectively. Throughout comprehensive evaluations on 7 datasets, including 4 maze datasets, 2 video-game datasets, and a real-world drone-view dataset containing 2 scenarios, we demonstrate remarkable improvements of our DAA* over neural A* in path similarity between the predicted and reference paths with a shorter path length when the shortest path is plausible, improving by 9.0% SPR, 6.9% ASIM, and 3.9% PSIM. Furthermore, when jointly learning pathfinding with both path loss and path probability map loss, DAA* significantly outperforms the state-of-the-art TransPath by 6.3% SPR, 6.0% PSIM, and 3.7% ASIM. We also discuss the minor trade-off between path optimality and search efficiency where applicable. Our code and model weights are available at https://github.com/zwxu064/DAAStar.git.
|
[
{
"version": "v1",
"created": "Sat, 12 Jul 2025 14:46:42 GMT"
},
{
"version": "v2",
"created": "Mon, 21 Jul 2025 08:36:50 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 06:51:41 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Xu",
"Zhiwei",
""
]
] |
2507.09623
|
Wenxing Zhang
|
Wenxing Zhang, Junle Pei, Xin Zhang, Tianjun Li
|
The First Constraint on the Dark Matter Millicharge via Multimessenger Time-Delay Analysis of GRB GW170817A
| null | null | null | null |
hep-ph
|
http://creativecommons.org/publicdomain/zero/1.0/
|
We derive new constraint on millicharged dark matter from the multimessenger observation of GW170817. In the neutron star merger event GW170817, the first detection of a gamma-ray burst (GRB) delayed by approximately 1.7 seconds relative to the gravitational wave emission was observed. Utilizing this delay, we constrain the millicharge parameter of dark matter within the large-scale structure of the Universe. {For dark matter mass below $10^{-15}$ eV, the millicharge parameter $\epsilon$ is constrained to be less than $10^{-14}$, representing the most stringent limits achieved to date.
|
[
{
"version": "v1",
"created": "Sun, 13 Jul 2025 13:17:18 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 12:13:47 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zhang",
"Wenxing",
""
],
[
"Pei",
"Junle",
""
],
[
"Zhang",
"Xin",
""
],
[
"Li",
"Tianjun",
""
]
] |
2507.09682
|
Armin Moin
|
Laura Baird and Armin Moin
|
OrQstrator: An AI-Powered Framework for Advanced Quantum Circuit Optimization
|
IEEE International Conference on Quantum Computing and Engineering (QCE) 2025 - Extended Abstract
| null | null | null |
cs.SE cs.AI cs.ET
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
We propose a novel approach, OrQstrator, which is a modular framework for conducting quantum circuit optimization in the Noisy Intermediate-Scale Quantum (NISQ) era. Our framework is powered by Deep Reinforcement Learning (DRL). Our orchestration engine intelligently selects among three complementary circuit optimizers: A DRL-based circuit rewriter trained to reduce depth and gate count via learned rewrite sequences; a domain-specific optimizer that performs efficient local gate resynthesis and numeric optimization; a parameterized circuit instantiator that improves compilation by optimizing template circuits during gate set translation. These modules are coordinated by a central orchestration engine that learns coordination policies based on circuit structure, hardware constraints, and backend-aware performance features such as gate count, depth, and expected fidelity. The system outputs an optimized circuit for hardware-aware transpilation and execution, leveraging techniques from an existing state-of-the-art approach, called the NISQ Analyzer, to adapt to backend constraints.
|
[
{
"version": "v1",
"created": "Sun, 13 Jul 2025 15:38:39 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 06:16:38 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Baird",
"Laura",
""
],
[
"Moin",
"Armin",
""
]
] |
2507.09712
|
Haoran Tang
|
Lingyi Chen and Haoran Tang and Shitong Wu and Jiakun Liu and Huihui Wu and Wenyi Zhang and Hao Wu
|
RDD Function: A Tradeoff Between Rate and Distortion-in-Distortion
| null | null | null | null |
cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this paper, we propose a novel function named Rate Distortion-in-Distortion (RDD) function as an extension of the classical rate-distortion (RD) function, where the expected distortion constraint is replaced by a Gromov-type distortion. This distortion, integral to the Gromov-Wasserstein (GW) distance, effectively defines the similarity in spaces of possibly different dimensions even without a direct metric between them. While the RDD function qualifies as an informational RD function, encoding theorems substantiate its status as an operational RD function, thereby underscoring its potential applicability in real-world source coding. Due to the high computational complexity associated with Gromov-type distortion, in general, the RDD function cannot be evaluated analytically. Consequently, we develop an alternating mirror descent algorithm that significantly reduces computational complexity by employing decomposition, linearization, and relaxation techniques. Numerical results on classical sources and different grids demonstrate the effectiveness of the developed algorithm. By exploring the relationship between the RDD function and the RD function, we suggest that the RDD function may have potential applications in future scenarios.
|
[
{
"version": "v1",
"created": "Sun, 13 Jul 2025 17:15:21 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 17:33:56 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Chen",
"Lingyi",
""
],
[
"Tang",
"Haoran",
""
],
[
"Wu",
"Shitong",
""
],
[
"Liu",
"Jiakun",
""
],
[
"Wu",
"Huihui",
""
],
[
"Zhang",
"Wenyi",
""
],
[
"Wu",
"Hao",
""
]
] |
2507.09871
|
Niket Patel
|
Niket Patel, Randall Balestriero
|
Task Priors: Enhancing Model Evaluation by Considering the Entire Space of Downstream Tasks
| null | null | null | null |
cs.LG cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
The grand goal of AI research, and particularly Self Supervised Learning (SSL), is to produce systems that can successfully solve any possible task. In contrast, current evaluation methods available to AI researchers typically rely on a fixed collection of hand-picked downstream benchmarks. Hence, a large amount of effort is put into designing and searching for large collection of evaluation tasks that can serve as a proxy of our grand goal. We argue that such a rigid evaluation protocol creates a silent bottleneck in AI research. To remedy that, we define a probabilistic space of downstream tasks obtained by adopting a distribution of tasks and by defining Task Priors. Under this view, one can evaluate a model's performance over the set of all possible downstream tasks. Our framework is the first to provide answers to key questions such as (i) what is the average performance of my model over all possible downstream tasks weighted by the probability to encounter each task? or (ii) what is the variance of my model's performance across all downstream tasks under the defined Task Priors? Beyond establishing a new standard for evaluation, we believe that Task Priors will accelerate the pace of research in SSL - where downstream task evaluation is the sole qualitative signal that researchers have access to.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 02:53:14 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 20:53:29 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Patel",
"Niket",
""
],
[
"Balestriero",
"Randall",
""
]
] |
2507.10055
|
Muhtadin Muhtadin
|
Muhtadin, I Wayan Agus Darmawan, Muhammad Hilmi Rusydiansyah, I Ketut Eddy Purnama, Chastine Fatichah, Mauridhi Hery Purnomo
|
Hand Gesture Recognition for Collaborative Robots Using Lightweight Deep Learning in Real-Time Robotic Systems
| null | null | null | null |
cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Direct and natural interaction is essential for intuitive human-robot collaboration, eliminating the need for additional devices such as joysticks, tablets, or wearable sensors. In this paper, we present a lightweight deep learning-based hand gesture recognition system that enables humans to control collaborative robots naturally and efficiently. This model recognizes eight distinct hand gestures with only 1,103 parameters and a compact size of 22 KB, achieving an accuracy of 93.5%. To further optimize the model for real-world deployment on edge devices, we applied quantization and pruning using TensorFlow Lite, reducing the final model size to just 7 KB. The system was successfully implemented and tested on a Universal Robot UR5 collaborative robot within a real-time robotic framework based on ROS2. The results demonstrate that even extremely lightweight models can deliver accurate and responsive hand gesture-based control for collaborative robots, opening new possibilities for natural human-robot interaction in constrained environments.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 08:40:24 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 04:25:41 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Muhtadin",
"",
""
],
[
"Darmawan",
"I Wayan Agus",
""
],
[
"Rusydiansyah",
"Muhammad Hilmi",
""
],
[
"Purnama",
"I Ketut Eddy",
""
],
[
"Fatichah",
"Chastine",
""
],
[
"Purnomo",
"Mauridhi Hery",
""
]
] |
2507.10062
|
Anil Koyuncu
|
Erg\"un Batuhan Kaynak, Mayasah Lami, Sahand Moslemi, Anil Koyuncu
|
LLMShot: Reducing snapshot testing maintenance via LLMs
|
Accepted to ICSME 2025
| null | null | null |
cs.SE
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Snapshot testing has emerged as a critical technique for UI validation in modern software development, yet it suffers from substantial maintenance overhead due to frequent UI changes causing test failures that require manual inspection to distinguish between genuine regressions and intentional design changes. This manual triage process becomes increasingly burdensome as applications evolve, creating a need for automated analysis solutions. This paper introduces LLMShot, a novel framework that leverages Vision-Language Models (VLMs) to automatically analyze snapshot test failures through semantic classification of UI changes. To evaluate LLMShot's effectiveness, we developed a comprehensive dataset using a feature-rich iOS application with configurable feature flags, creating realistic scenarios that produce authentic snapshot differences representative of real development workflows. Our evaluation using Gemma3 models demonstrates strong classification performance, with the 12B variant achieving over 84% recall in identifying failure root causes while the 4B model offers practical deployment advantages with acceptable performance for continuous integration environments. However, our exploration of selective ignore mechanisms revealed significant limitations in current prompting-based approaches for controllable visual reasoning. LLMShot represents the first automated approach to semantic snapshot test analysis, offering developers structured insights that can substantially reduce manual triage effort and advance toward more intelligent UI testing paradigms.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 08:47:19 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 11:58:01 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kaynak",
"Ergün Batuhan",
""
],
[
"Lami",
"Mayasah",
""
],
[
"Moslemi",
"Sahand",
""
],
[
"Koyuncu",
"Anil",
""
]
] |
2507.10084
|
Xin Tong
|
Haonan Chen (Tibet University) and Xin Tong (Northwestern Polytechnical University)
|
A Transfer Learning-Based Method for Water Body Segmentation in Remote Sensing Imagery: A Case Study of the Zhada Tulin Area
|
13 pages, 6 figures, 2 tables
| null | null | null |
cs.CV cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The Tibetan Plateau, known as the Asian Water Tower, faces significant water security challenges due to its high sensitivity to climate change. Advancing Earth observation for sustainable water monitoring is thus essential for building climate resilience in this region. This study proposes a two-stage transfer learning strategy using the SegFormer model to overcome domain shift and data scarcit--key barriers in developing robust AI for climate-sensitive applications. After pre-training on a diverse source domain, our model was fine-tuned for the arid Zhada Tulin area. Experimental results show a substantial performance boost: the Intersection over Union (IoU) for water body segmentation surged from 25.50% (direct transfer) to 64.84%. This AI-driven accuracy is crucial for disaster risk reduction, particularly in monitoring flash flood-prone systems. More importantly, the high-precision map reveals a highly concentrated spatial distribution of water, with over 80% of the water area confined to less than 20% of the river channel length. This quantitative finding provides crucial evidence for understanding hydrological processes and designing targeted water management and climate adaptation strategies. Our work thus demonstrates an effective technical solution for monitoring arid plateau regions and contributes to advancing AI-powered Earth observation for disaster preparedness in critical transboundary river headwaters.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 09:11:33 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:37:18 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Chen",
"Haonan",
"",
"Tibet University"
],
[
"Tong",
"Xin",
"",
"Northwestern Polytechnical University"
]
] |
2507.10136
|
Zhonglin Liu
|
Zhonglin Liu
|
A PBN-RL-XAI Framework for Discovering a "Hit-and-Run" Therapeutic Strategy in Melanoma
|
9 pages, 5 figures. Submitted to the IEEE International Conference on Bioinformatics and Biomedicine (BIBM) 2025. Code is available at https://github.com/Liu-Zhonglin/pbn-melanoma-project
| null | null | null |
q-bio.QM cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Innate resistance to anti-PD-1 immunotherapy remains a major clinical challenge in metastatic melanoma, with the underlying molecular networks being poorly understood. To address this, we constructed a dynamic Probabilistic Boolean Network model using transcriptomic data from patient tumor biopsies to elucidate the regulatory logic governing therapy response. We then employed a reinforcement learning agent to systematically discover optimal, multi-step therapeutic interventions and used explainable artificial intelligence to mechanistically interpret the agent's control policy. The analysis revealed that a precisely timed, 4-step temporary inhibition of the lysyl oxidase like 2 protein (LOXL2) was the most effective strategy. Our explainable analysis showed that this ''hit-and-run" intervention is sufficient to erase the molecular signature driving resistance, allowing the network to self-correct without requiring sustained intervention. This study presents a novel, time-dependent therapeutic hypothesis for overcoming immunotherapy resistance and provides a powerful computational framework for identifying non-obvious intervention protocols in complex biological systems.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 10:35:38 GMT"
},
{
"version": "v2",
"created": "Tue, 15 Jul 2025 01:37:39 GMT"
},
{
"version": "v3",
"created": "Wed, 16 Jul 2025 13:21:18 GMT"
},
{
"version": "v4",
"created": "Thu, 24 Jul 2025 04:04:13 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Liu",
"Zhonglin",
""
]
] |
2507.10214
|
Srijan Sehgal Dr.
|
Srijan Sehgal (for the Pierre Auger Collaboration)
|
Search for Ultra-High-Energy Neutrinos at the Pierre Auger Observatory: New Triggers, Methods, and Constraints
|
Presented at the 39th International Cosmic Ray Conference (ICRC 2025). 13 pages, 5 figures
| null | null |
PoS-ICRC2025-1170
|
astro-ph.HE astro-ph.IM
|
http://creativecommons.org/licenses/by-sa/4.0/
|
The Pierre Auger Observatory has the capability to identify neutrino-induced extensive air showers above $10^{17}$ eV by using its large Surface Detector (SD) array. Data from the Observatory have been used to set some of the most stringent upper limits to the neutrino flux in the ultra-high energy (UHE) range. The data have also been used for follow-up detection of transient events in the context of multi-messenger astrophysics. In mid-2013, two additional SD triggers (Time-over-Threshold-deconvolved (ToTd) and Multiplicity-of-Positive Steps (MoPS)) were shown to increase the detection capability for the neutrino-induced air showers in the energy regime below $10^{19}$ eV by a factor of 5-10. This contribution will give an overview of the ongoing work regarding the searches for UHE neutrinos at the Pierre Auger Observatory. The impact of the ToTd and MoPS triggers for neutrino search in the zenith angle range of $60^{\circ} < \theta < 75^{\circ}$ is discussed. A novel neutrino identification method, which integrates these triggers, is applied to observational data to look for neutrino-like events using a $\textit{blind}$ search strategy. New constraints to point-like sources of UHE neutrinos will be presented for the angular range explored.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 12:31:42 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 11:20:15 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Sehgal",
"Srijan",
"",
"for the Pierre Auger Collaboration"
]
] |
2507.10244
|
Adam \v{S}t\v{e}p\'anek
|
Adam \v{S}t\v{e}p\'anek, David Ku\v{t}\'ak, Barbora Kozl\'ikov\'a, Jan By\v{s}ka
|
Helveg: Diagrams for Software Documentation
|
13 pages, 5 figures, accepted by TVCG
| null |
10.1109/TVCG.2025.3589748
| null |
cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Software developers often have to gain an understanding of a codebase. Be it programmers getting onboarded onto a team project or, for example, developers striving to grasp an external open-source library. In either case, they frequently turn to the project's documentation. However, documentation in its traditional textual form is ill-suited for this kind of high-level exploratory analysis, since it is immutable from the readers' perspective and thus forces them to follow a predefined path. We have designed an approach bringing aspects of software architecture visualization to API reference documentation. It utilizes a highly interactive node-link diagram with expressive node glyphs and flexible filtering capabilities, providing a high-level overview of the codebase as well as details on demand. To test our design, we have implemented a prototype named Helveg, capable of automatically generating diagrams of C\# codebases. User testing of Helveg confirmed its potential, but it also revealed problems with the readability, intuitiveness, and user experience of our tool. Therefore, in this paper, which is an extended version of our VISSOFT paper with DOI 10.1109/VISSOFT64034.2024.00012, we address many of these problems through major changes to the glyph design, means of interaction, and user interface of the tool. To assess the improvements, this new version of Helveg was evaluated again with the same group of participants as the previous version.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 13:07:50 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Štěpánek",
"Adam",
""
],
[
"Kuťák",
"David",
""
],
[
"Kozlíková",
"Barbora",
""
],
[
"Byška",
"Jan",
""
]
] |
2507.10351
|
Dieter Rautenbach
|
Dieter Rautenbach and Johannes Scherer and Florian Werner
|
Leaf to leaf path lengths in trees of given degree sequence
|
6 pages, 0 figures
| null | null | null |
math.CO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
For a tree $T$, let $lp(T)$ be the number of different lengths of leaf to leaf paths in $T$. For a degree sequence $s$ of a tree, let ${\rm rad}(s)$ be the minimum radius of a tree with degree sequence $s$. Recently, Di Braccio, Katsamaktsis, Ma, Malekshahian, and Zhao provided a lower bound on $lp(T)$ in terms of the number of leaves and the maximum degree of $T$, answering a related question posed by Narins, Pokrovskiy, and Szab\'o. Here we show $lp(T)\geq {\rm rad}(s)-\log_2\left({\rm rad}(s)\right)$ for a tree $T$ with no vertex of degree $2$ and degree sequence $s$, and discuss possible improvements and variants.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 14:54:10 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 05:36:35 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Rautenbach",
"Dieter",
""
],
[
"Scherer",
"Johannes",
""
],
[
"Werner",
"Florian",
""
]
] |
2507.10384
|
Hugo Roussille
|
Jibril Ben Achour, Clara Montagnon, Hugo Roussille
|
Algebraically special perturbations of the Kerr black hole: a metric formulation
|
22+3 pages, 1 figure
| null | null | null |
gr-qc hep-th
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Perturbations of the Kerr black hole are notoriously difficult to describe in the metric formalism and are usually studied in terms of perturbations of the Weyl scalars. In this work, we focus on the algebraically special linear perturbations (ASLP) of the Kerr geometry and show how one can describe this subsector of the perturbations solely using the metric formulation. To that end, we consider the most general twisting algebraically special solution space of vacuum General Relativity. By linearizing around the Kerr solution, we obtain two coupled partial differential wave equations describing the dynamics of the Kerr ASLP. We provide an algorithm to solve them analytically in the small spin approximation up to third order, providing the first exact solution of this kind in the metric formulation. Then, we use this framework to study the stationary zero modes of the Kerr geometry. We present the exact analytical form of the shifts in mass and spin together with the required change of coordinates needed to identify them. Finally, we also provide for the first time closed expressions for the solution-generating perturbations generating the NUT and acceleration charges, thus deforming the Kerr solution to the linearized Kerr-NUT and spinning C-metric. These results provide a first concrete and rare example of perturbations of the Kerr black hole which can be treated entirely in the metric formulation. They can serve as a useful testbed to search for hidden symmetries of the Kerr perturbations.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 15:23:27 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:29:37 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Achour",
"Jibril Ben",
""
],
[
"Montagnon",
"Clara",
""
],
[
"Roussille",
"Hugo",
""
]
] |
2507.10476
|
Alexander Boyd
|
Kyle J. Ray and Alexander B. Boyd
|
Large Interconnected Thermodynamic Systems Nearly Minimize Entropy Production
| null | null | null | null |
cond-mat.stat-mech
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Many have speculated whether nonequilibrium systems obey principles of maximum or minimum entropy production. In this work, we use stochastic thermodynamics to derive the condition for the minimum entropy production state (MEPS) for continuous-time Markov chains (CTMCs), even far from equilibrium. We show that real nonequilibrium steady states (NESS) generally violate both the MINEP and MAXEP principles. However, through numerical sampling of large interconnected CTMCs, we find that as system size increases, the steady-state entropy production tends to converge toward the minimum. This suggests that large nonequilibrium systems may self-organize to make efficient use of thermodynamic resources, offering a nuanced perspective on the longstanding debate between MAXEP and MINEP.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 16:56:04 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 21:23:48 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Ray",
"Kyle J.",
""
],
[
"Boyd",
"Alexander B.",
""
]
] |
2507.10732
|
Josée Desharnais
|
Jos\'ee Desharnais and Ana Sokolova
|
$\epsilon$-Distance via L\'evy-Prokhorov Lifting
| null | null | null | null |
cs.LO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The most studied and accepted pseudometric for probabilistic processes is one based on the Kantorovich distance between distributions. It comes with many theoretical and motivating results, in particular it is the fixpoint of a given functional and defines a functor on (complete) pseudometric spaces.
Other notions of behavioural pseudometrics have also been proposed, one of them ($\epsilon$-distance) based on $\epsilon$-bisimulation. $\epsilon$-Distance has the advantages that it is intuitively easy to understand, it relates systems that are conceptually close (for example, an imperfect implementation is close to its specification), and it comes equipped with a natural notion of $\epsilon$-coupling. Finally, this distance is easy to compute.
We show that $\epsilon$-distance is also the greatest fixpoint of a functional and provides a functor. The latter is obtained by replacing the Kantorovich distance in the lifting functor with the L\'evy-Prokhorov distance. In addition, we show that $\epsilon$-couplings and $\epsilon$-bisimulations have an appealing coalgebraic characterization.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 18:54:51 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 14:20:02 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Desharnais",
"Josée",
""
],
[
"Sokolova",
"Ana",
""
]
] |
2507.10806
|
Esteban Roulet
|
Esteban Roulet (for the Pierre Auger Collaboration)
|
Combined fit above 0.1 EeV to the cosmic-ray spectrum and composition measured with the Pierre Auger Observatory
|
Presented at the 39th International Cosmic Ray Conference (ICRC 2025)
| null | null |
PoS-ICRC2025-373
|
astro-ph.HE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We fit the cosmic-ray spectrum measured with the Pierre Auger Observatory's surface detectors above an energy of $10^{17}$ eV, along with composition information inferred from the depth of shower maximum measured with its fluorescence detectors above an energy threshold of $10^{17.8}$ eV. We consider astrophysical scenarios with two distinct extragalactic source populations: one dominating the flux above a few EeV, and the other dominating at lower energies, with representative nuclei being injected at the sources with power-law spectra and rigidity-dependent cutoffs. The high-energy population exhibits a hard source injection spectrum and a relatively heavy composition, while the low-energy population exhibits a softer spectrum and a lighter composition. Extending the fit down to the low energies considered here shows the potential to test the energy region towards the Galactic to extragalactic transition with the data of the Pierre Auger Observatory. In particular, the Galactic contribution is expected to be still sizable at the lowest energies considered, and the extragalactic contribution needs to become suppressed for decreasing energies, an effect that could naturally result from a magnetic-horizon-induced suppression.
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 20:59:10 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Roulet",
"Esteban",
"",
"for the Pierre Auger Collaboration"
]
] |
2507.11120
|
Ranjith R Kumar
|
Ranjith R Kumar and Pasquale Marra
|
Topologically nontrivial multicritical points
|
12 pages, 13 figures
| null | null | null |
cond-mat.dis-nn cond-mat.mes-hall cond-mat.str-el hep-th
|
http://creativecommons.org/licenses/by/4.0/
|
Recently, the intriguing interplay between topology and quantum criticality has been unveiled in one-dimensional topological chains with extended nearest-neighbor couplings. In these systems, topologically distinct critical phases emerge with localized edge modes despite the vanishing bulk gap. In this work, we study the topological multicritical points at which distinct gapped and critical phases intersect. Specifically, we consider a topological chain with coupling up to the third nearest neighbors, which shows stable localized edge modes at the multicritical points. These points possess only nontrivial gapped and critical phases around them and are also characterized by the quadratic dispersion around the gap-closing points. We characterize the topological multicritical points in terms of the topological invariant obtained from the zeros of the complex function associated with the Hamiltonian. Further, we analyze the nature of zeros in the vicinity of the multicritical points by calculating the discriminants of the associated polynomial. The discriminant uniquely identifies the topological multicritical points and distinguishes them from the trivial ones. We finally study the robustness of the zero-energy modes at the multicritical points at weak disorder strengths, and reveal the presence of a topologically nontrivial gapless Anderson-localized phase at strong disorder strengths.
|
[
{
"version": "v1",
"created": "Tue, 15 Jul 2025 09:14:20 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kumar",
"Ranjith R",
""
],
[
"Marra",
"Pasquale",
""
]
] |
2507.11361
|
Maximilian Bernecker
|
Maximilian Bernecker, Smaranda Sgarciu, Xiaoming Kan, Mehrnaz Anvari, Iegor Riepin, Felix M\"usgens
|
Adaptive Robust Optimization for European Electricity System Planning Considering Regional Dunkelflaute Events
|
Code and data can be found on github: https://github.com/bernemax/ARO_Dunkelflaute_Europe
| null | null | null |
econ.GN q-fin.EC
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
This study develops a capacity expansion model for a fully decarbonized European electricity system using an Adaptive Robust Optimization (ARO) framework. The model endogenously identifies the worst regional Dunkelflaute events, prolonged periods of low wind and solar availability, and incorporates multiple extreme weather realizations within a single optimization run. Results show that system costs rise nonlinearly with the geographic extent of these events: a single worst-case regional disruption increases costs by 9%, but broader disruptions across multiple regions lead to much sharper increases, up to 51%. As Dunkelflaute conditions extend across most of Europe, additional cost impacts level off, with a maximum increase of 71%. The optimal technology mix evolves with the severity of weather stress: while renewables, batteries, and interregional transmission are sufficient to manage localized events, large-scale disruptions require long-term hydrogen storage and load shedding to maintain system resilience. Central European regions, especially Germany and France, emerge as systemic bottlenecks, while peripheral regions bear the cost of compensatory overbuilding. These findings underscore the need for a coordinated European policy strategy that goes beyond national planning to support cross-border infrastructure investment, scale up flexible technologies such as long-duration storage, and promote a geographically balanced deployment of renewables to mitigate systemic risks associated with Dunkelflaute events.
|
[
{
"version": "v1",
"created": "Tue, 15 Jul 2025 14:28:23 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 10:05:06 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Bernecker",
"Maximilian",
""
],
[
"Sgarciu",
"Smaranda",
""
],
[
"Kan",
"Xiaoming",
""
],
[
"Anvari",
"Mehrnaz",
""
],
[
"Riepin",
"Iegor",
""
],
[
"Müsgens",
"Felix",
""
]
] |
2507.11554
|
Yize Li
|
Zejian Li, Yize Li, Chenye Meng, Zhongni Liu, Yang Ling, Shengyuan Zhang, Guang Yang, Changyuan Yang, Zhiyuan Yang, Lingyun Sun
|
Inversion-DPO: Precise and Efficient Post-Training for Diffusion Models
|
Accepted by ACM MM25
| null | null | null |
cs.CV cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent advancements in diffusion models (DMs) have been propelled by alignment methods that post-train models to better conform to human preferences. However, these approaches typically require computation-intensive training of a base model and a reward model, which not only incurs substantial computational overhead but may also compromise model accuracy and training efficiency. To address these limitations, we propose Inversion-DPO, a novel alignment framework that circumvents reward modeling by reformulating Direct Preference Optimization (DPO) with DDIM inversion for DMs. Our method conducts intractable posterior sampling in Diffusion-DPO with the deterministic inversion from winning and losing samples to noise and thus derive a new post-training paradigm. This paradigm eliminates the need for auxiliary reward models or inaccurate appromixation, significantly enhancing both precision and efficiency of training. We apply Inversion-DPO to a basic task of text-to-image generation and a challenging task of compositional image generation. Extensive experiments show substantial performance improvements achieved by Inversion-DPO compared to existing post-training methods and highlight the ability of the trained generative models to generate high-fidelity compositionally coherent images. For the post-training of compostitional image geneation, we curate a paired dataset consisting of 11,140 images with complex structural annotations and comprehensive scores, designed to enhance the compositional capabilities of generative models. Inversion-DPO explores a new avenue for efficient, high-precision alignment in diffusion models, advancing their applicability to complex realistic generation tasks. Our code is available at https://github.com/MIGHTYEZ/Inversion-DPO
|
[
{
"version": "v1",
"created": "Mon, 14 Jul 2025 02:59:28 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Jul 2025 09:25:36 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 10:37:32 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Li",
"Zejian",
""
],
[
"Li",
"Yize",
""
],
[
"Meng",
"Chenye",
""
],
[
"Liu",
"Zhongni",
""
],
[
"Ling",
"Yang",
""
],
[
"Zhang",
"Shengyuan",
""
],
[
"Yang",
"Guang",
""
],
[
"Yang",
"Changyuan",
""
],
[
"Yang",
"Zhiyuan",
""
],
[
"Sun",
"Lingyun",
""
]
] |
2507.11715
|
Rafael Azuaje
|
Rafael Azuaje and Piergiulio Tempesta
|
Jacobi-Haantjes manifolds, integrability and dissipation
| null | null | null | null |
math-ph math.DG math.MP
|
http://creativecommons.org/licenses/by/4.0/
|
The theory of Jacobi-Haantjes manifolds is proposed as a natural geometric framework useful to state the integrability of both conservative and dissipative Hamiltonian systems in a unified way. As a reduction, new structures such as contact-Haantjes and locally conformal symplectic-Haantjes manifolds are investigated. The main properties of these novel classes of manifolds and their mutual relationship are discussed. It is shown that the notion of Haantjes chains, within the framework of Jacobi geometry, is related to the existence of complete integrable contact Hamiltonian systems, encompassing also dissipative cases, as well as to the integrability of Hamiltonian systems on appropriate invariant submanifolds. Our framework naturally generalizes the formulation of integrable conservative Hamiltonian systems through the theory of symplectic-Haantjes (or {\omega}H) manifolds.
|
[
{
"version": "v1",
"created": "Tue, 15 Jul 2025 20:32:20 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 00:30:47 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Azuaje",
"Rafael",
""
],
[
"Tempesta",
"Piergiulio",
""
]
] |
2507.11783
|
Neeraj Wagh
|
Gayal Kuruppu, Neeraj Wagh, Yogatheesan Varatharajah
|
EEG Foundation Models: A Critical Review of Current Progress and Future Directions
|
20 pages, 5 figures, 3 tables (main + supplement)
| null | null | null |
eess.SP cs.AI cs.LG q-bio.NC
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Patterns of electrical brain activity recorded via electroencephalography (EEG) offer immense value for scientific and clinical investigations. The inability of supervised EEG encoders to learn robust EEG patterns and their over-reliance on expensive signal annotations have sparked a transition towards general-purpose self-supervised EEG encoders, i.e., EEG foundation models (EEG-FMs), for robust and scalable EEG feature extraction. However, the real-world readiness of early EEG-FMs and the rubric for long-term research progress remain unclear. A systematic and comprehensive review of first-generation EEG-FMs is therefore necessary to understand the current state-of-the-art and identify key directions for future EEG-FMs. To that end, this study reviews 10 early EEG-FMs and presents a critical synthesis of their methodology, empirical findings, and outstanding research gaps. We find that most EEG-FMs adopt a sequence-based modeling scheme that relies on transformer-based backbones and the reconstruction of masked sequences for self-supervision. However, model evaluations remain heterogeneous and largely limited, making it challenging to assess their practical off-the-shelf utility. In addition to adopting standardized and realistic evaluations, future work should demonstrate more substantial scaling effects and make principled and trustworthy choices throughout the EEG representation learning pipeline. We believe that developing benchmarks, software tools, technical methodologies, and applications in collaboration with domain experts may further advance the translational utility and real-world adoption of EEG-FMs.
|
[
{
"version": "v1",
"created": "Tue, 15 Jul 2025 22:52:44 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 20:10:43 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kuruppu",
"Gayal",
""
],
[
"Wagh",
"Neeraj",
""
],
[
"Varatharajah",
"Yogatheesan",
""
]
] |
2507.11936
|
Jianzhe Ma
|
Jianzhe Ma, Wenxuan Wang, Qin Jin
|
A Survey of Deep Learning for Geometry Problem Solving
|
Work in progress
| null | null | null |
cs.CL cs.AI cs.CV cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Geometry problem solving is a key area of mathematical reasoning, which is widely involved in many important fields such as education, mathematical ability assessment of artificial intelligence, and multimodal ability assessment. In recent years, the rapid development of deep learning technology, especially the rise of multimodal large language models, has triggered a widespread research boom. This paper provides a survey of the applications of deep learning in geometry problem solving, including (i) a comprehensive summary of the relevant tasks in geometry problem solving; (ii) a thorough review of related deep learning methods; (iii) a detailed analysis of evaluation metrics and methods; and (iv) a critical discussion of the current challenges and future directions that can be explored. Our goal is to provide a comprehensive and practical reference of deep learning for geometry problem solving to promote further developments in this field. We create a continuously updated list of papers on GitHub: https://github.com/majianz/dl4gps.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 06:03:08 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Jul 2025 12:18:01 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 06:15:29 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Ma",
"Jianzhe",
""
],
[
"Wang",
"Wenxuan",
""
],
[
"Jin",
"Qin",
""
]
] |
2507.11983
|
Ngoc Duc Le
|
Ngoc Duc Le, D.-H.-Minh Nguyen, Dung Xuan Nguyen, Hai Son Nguyen, Dario Bercioux
|
Berry Monopole Scattering in the Synthetic Momentum Space of a Bilayer Photonic Crystal Slab
|
46 pages, 21 figures
| null | null | null |
physics.optics cond-mat.mes-hall
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Berry monopoles-quantized sources of Berry curvature-are fundamental to topological phases, yet their scattering remains unexplored. Here, we report for the first time the adiabatic scattering of Berry monopoles in a bilayer photonic crystal slab combining one genuine and one synthetic momentum. Two monopoles approach, collide, and scatter within this hybrid parameter space. The process is described by an effective coupled-mode model and confirmed by full-wave simulations. We further propose an experimental scheme using chiral edge states, opening a route to probe monopole interactions in synthetic photonic systems.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 07:29:18 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 07:13:14 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Le",
"Ngoc Duc",
""
],
[
"Nguyen",
"D. -H. -Minh",
""
],
[
"Nguyen",
"Dung Xuan",
""
],
[
"Nguyen",
"Hai Son",
""
],
[
"Bercioux",
"Dario",
""
]
] |
2507.12006
|
Linwei Chen
|
Linwei Chen, Lin Gu, Ying Fu
|
Frequency-Dynamic Attention Modulation for Dense Prediction
|
Accepted by ICCV 2025
| null | null | null |
cs.CV cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Vision Transformers (ViTs) have significantly advanced computer vision, demonstrating strong performance across various tasks. However, the attention mechanism in ViTs makes each layer function as a low-pass filter, and the stacked-layer architecture in existing transformers suffers from frequency vanishing. This leads to the loss of critical details and textures. We propose a novel, circuit-theory-inspired strategy called Frequency-Dynamic Attention Modulation (FDAM), which can be easily plugged into ViTs. FDAM directly modulates the overall frequency response of ViTs and consists of two techniques: Attention Inversion (AttInv) and Frequency Dynamic Scaling (FreqScale). Since circuit theory uses low-pass filters as fundamental elements, we introduce AttInv, a method that generates complementary high-pass filtering by inverting the low-pass filter in the attention matrix, and dynamically combining the two. We further design FreqScale to weight different frequency components for fine-grained adjustments to the target response function. Through feature similarity analysis and effective rank evaluation, we demonstrate that our approach avoids representation collapse, leading to consistent performance improvements across various models, including SegFormer, DeiT, and MaskDINO. These improvements are evident in tasks such as semantic segmentation, object detection, and instance segmentation. Additionally, we apply our method to remote sensing detection, achieving state-of-the-art results in single-scale settings. The code is available at https://github.com/Linwei-Chen/FDAM.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 07:59:54 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 03:10:31 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 09:57:56 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Chen",
"Linwei",
""
],
[
"Gu",
"Lin",
""
],
[
"Fu",
"Ying",
""
]
] |
2507.12068
|
Mohammad Javad Habibi Vosta Kolaei
|
Mohammad Javad Habibi Vosta Kolaei
|
Analyzing the Geometry of Immersions of Co-Dimension One via Shape Operator Dynamics
|
24 pages
| null | null | null |
math.DG
|
http://creativecommons.org/licenses/by/4.0/
|
We study the extrinsic geometry of isometric immersions into Riemannian manifolds of co-dimension one via a fourth-order geometric evolution of the shape operator. Motivated by bi-harmonic map theory and generalized Chen's conjecture, we introduce a moduli flow, a tensorial gradient flow that decreases a natural energy measuring curvature variation.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 09:24:35 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 07:14:43 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kolaei",
"Mohammad Javad Habibi Vosta",
""
]
] |
2507.12106
|
Antonio Salis
|
Antonio Salis, Gabriele Troina, Gianluca Boanelli, Marco Ottaviano, Paola Fortini, Soraya Versace
|
Urban Green Governance: IoT-Driven Management and Enhancement of Urban Green Spaces in Campobasso
|
18 pages, 6 Figures
| null | null | null |
cs.DC cs.CY
|
http://creativecommons.org/licenses/by/4.0/
|
The efficient design and management of public green spaces is a key factor in promoting the health and well-being of urban population, as emphasized by the WHO, UNEP, and EEA. These areas serve as the "green lungs" of the urban ecosystem, playing a vital role in enhancing quality of life thanks to the provision of ecosystem services. In this context, the Smart Green City use case in Campobasso municipality, funded by the Italian Ministry of Enterprises (MIMIT), emerges as an innovative model for the sustainable management of green urban areas through the adoption of an advanced system of emerging technologies integrated and interoperable. The project integrates IoT systems and data-driven governance platforms, enabling real-time monitoring of the health status of trees and green areas via a Decision Support System (DSS). It also facilitates the collection and analysis of data from diverse sources, including weather conditions, air quality, soil moisture, pollution levels. The resulting cloud-based platform supports a holistic real time decision making for green urban managers, technical experts and operational staff. It enables intelligent control and management of urban green spaces using Tree Talker sensors, integrated with soil moisture and water potential monitoring systems. Thanks to predictive models based on machine learning algorithms and real time data provided by IoT sensors, irrigation of public parks can be optimized by providing suggestions on when and how much water to apply. Customized alerts layers are also activated warning users when monitored parameters, such as soil temperature, humidity, or water potential, exceed predefined thresholds. This Use Case demonstrates how digitalization, IoT sensors fusion and technological innovation can support sustainable urban governance, fostering environmental resilience and improving citizens quality of life.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 10:24:07 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 15:06:54 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 14:45:04 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Salis",
"Antonio",
""
],
[
"Troina",
"Gabriele",
""
],
[
"Boanelli",
"Gianluca",
""
],
[
"Ottaviano",
"Marco",
""
],
[
"Fortini",
"Paola",
""
],
[
"Versace",
"Soraya",
""
]
] |
2507.12311
|
Zhangcheng Qiang
|
Zhangcheng Qiang
|
An Ecosystem for Ontology Interoperability
|
5 pages, 8 figures
| null | null | null |
cs.IR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Ontology interoperability is one of the complicated issues that restricts the use of ontologies in knowledge graphs (KGs). Different ontologies with conflicting and overlapping concepts make it difficult to design, develop, and deploy an interoperable ontology for downstream tasks. We propose an ecosystem for ontology interoperability. The ecosystem employs three state-of-the-art semantic techniques in different phases of the ontology engineering life cycle: ontology design patterns (ODPs) in the design phase, ontology matching and versioning (OM\&OV) in the develop phase, and ontology-compliant knowledge graphs (OCKGs) in the deploy phase, to achieve better ontology interoperability in real-world applications. A case study of sensor observation in the building domain validates the usefulness of the proposed ecosystem.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 15:06:29 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 18:19:49 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Qiang",
"Zhangcheng",
""
]
] |
2507.12558
|
Phuong Nguyen
|
Tien P. T. Le, Anh M. T. Bui, Huy N. D. Pham, Alessio Bucaioni, Phuong T. Nguyen
|
When Retriever Meets Generator: A Joint Model for Code Comment Generation
|
The paper has been peer-reviewed and accepted for publication in the proceedings of the 19th ACM/IEEE International Symposium on Empirical Software Engineering and Measurement (ESEM 2025)
| null | null | null |
cs.SE
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Automatically generating concise, informative comments for source code can lighten documentation effort and accelerate program comprehension. Retrieval-augmented approaches first fetch code snippets with existing comments and then synthesize a new comment, yet retrieval and generation are typically optimized in isolation, allowing irrelevant neighbors topropagate noise downstream. To tackle the issue, we propose a novel approach named RAGSum with the aim of both effectiveness and efficiency in recommendations. RAGSum is built on top offuse retrieval and generation using a single CodeT5 backbone. We report preliminary results on a unified retrieval-generation framework built on CodeT5. A contrastive pre-training phase shapes code embeddings for nearest-neighbor search; these weights then seed end-to-end training with a composite loss that (i) rewards accurate top-k retrieval; and (ii) minimizes comment-generation error. More importantly, a lightweight self-refinement loop is deployed to polish the final output. We evaluated theframework on three cross-language benchmarks (Java, Python, C), and compared it with three well-established baselines. The results show that our approach substantially outperforms thebaselines with respect to BLEU, METEOR, and ROUTE-L. These findings indicate that tightly coupling retrieval and generationcan raise the ceiling for comment automation and motivateforthcoming replications and qualitative developer studies.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 18:12:27 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 07:07:10 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Le",
"Tien P. T.",
""
],
[
"Bui",
"Anh M. T.",
""
],
[
"Pham",
"Huy N. D.",
""
],
[
"Bucaioni",
"Alessio",
""
],
[
"Nguyen",
"Phuong T.",
""
]
] |
2507.12570
|
Subhadip Bisal
|
Subhadip Bisal
|
Constraining ALP-Top Interaction from the Chromoelectric Dipole Moment of the Top Quark
|
26 pages, 4 figures
| null | null | null |
hep-ph
|
http://creativecommons.org/licenses/by/4.0/
|
The couplings of axion-like particles (ALPs) to Standard Model fermions are proportional to the fermion masses, making the interaction with the top quark particularly significant. In this study, we consider an ALP that is a mixture of CP-even and CP-odd components, thereby introducing CP violation. This CP violation, in turn, gives rise to electric dipole moments (EDMs) of quarks and leptons, as well as chromoelectric dipole moments (CEDMs) of quarks. We compute the one-loop and two-loop contributions to the top quark CEDM induced by the ALP. In our calculation, we treat the external gluon as off-shell with momentum $q^2 \neq 0$, derive the analytical results, and finally evaluate the top quark CEDM at $q^2 = m_t^2$, corresponding to the top quark pole mass. This value is relevant for subsequent calculations of the EDMs of the neutron and mercury. By applying current experimental bounds on EDMs and CEDMs, we derive constraints on the ALP-top quark coupling, with the strongest limit coming from the neutron EDM.
|
[
{
"version": "v1",
"created": "Wed, 16 Jul 2025 18:38:35 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 08:48:26 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Bisal",
"Subhadip",
""
]
] |
2507.12720
|
Abraham Toluwase Owodunni
|
Abraham Toluase Owodunni, Orevaoghene Ahia, Sachin Kumar
|
FLEXITOKENS: Flexible Tokenization for Evolving Language Models
| null | null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Language models (LMs) are challenging to adapt to new data distributions by simple finetuning. This is due to the rigidity of their subword tokenizers, which typically remain unchanged during adaptation. This inflexibility often leads to inefficient tokenization, causing overfragmentation of out-of-distribution domains, unseen languages, or scripts. In this work, we develop byte-level LMs with learnable tokenizers to make tokenization adaptive. Our models include a submodule that learns to predict boundaries between the input byte sequence, encoding it into variable-length segments. Existing tokenizer-free methods train this boundary predictor using an auxiliary loss that enforces a fixed compression rate across the training corpus, introducing a new kind of rigidity. We propose FLEXITOKENS, a simplified training objective that enables significantly greater flexibility during adaptation. Evaluating across multiple multilingual benchmarks, morphologically diverse tasks, and domains, we demonstrate that FLEXITOKENS consistently reduces token over-fragmentation and achieves up to 10% improvements on downstream task performance compared to subword and other gradient-based tokenizers. Code and data for our experiments will be released at https://github.com/owos/flexitokens
|
[
{
"version": "v1",
"created": "Thu, 17 Jul 2025 01:55:41 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 18:40:42 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Owodunni",
"Abraham Toluase",
""
],
[
"Ahia",
"Orevaoghene",
""
],
[
"Kumar",
"Sachin",
""
]
] |
2507.12890
|
Huakang Chen
|
Huakang Chen, Yuepeng Jiang, Guobin Ma, Chunbo Hao, Shuai Wang, Jixun Yao, Ziqian Ning, Meng Meng, Jian Luan, Lei Xie
|
DiffRhythm+: Controllable and Flexible Full-Length Song Generation with Preference Optimization
| null | null | null | null |
eess.AS cs.SD
|
http://creativecommons.org/licenses/by/4.0/
|
Songs, as a central form of musical art, exemplify the richness of human intelligence and creativity. While recent advances in generative modeling have enabled notable progress in long-form song generation, current systems for full-length song synthesis still face major challenges, including data imbalance, insufficient controllability, and inconsistent musical quality. DiffRhythm, a pioneering diffusion-based model, advanced the field by generating full-length songs with expressive vocals and accompaniment. However, its performance was constrained by an unbalanced model training dataset and limited controllability over musical style, resulting in noticeable quality disparities and restricted creative flexibility. To address these limitations, we propose DiffRhythm+, an enhanced diffusion-based framework for controllable and flexible full-length song generation. DiffRhythm+ leverages a substantially expanded and balanced training dataset to mitigate issues such as repetition and omission of lyrics, while also fostering the emergence of richer musical skills and expressiveness. The framework introduces a multi-modal style conditioning strategy, enabling users to precisely specify musical styles through both descriptive text and reference audio, thereby significantly enhancing creative control and diversity. We further introduce direct performance optimization aligned with user preferences, guiding the model toward consistently preferred outputs across evaluation metrics. Extensive experiments demonstrate that DiffRhythm+ achieves significant improvements in naturalness, arrangement complexity, and listener satisfaction over previous systems.
|
[
{
"version": "v1",
"created": "Thu, 17 Jul 2025 08:18:55 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 08:15:02 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Chen",
"Huakang",
""
],
[
"Jiang",
"Yuepeng",
""
],
[
"Ma",
"Guobin",
""
],
[
"Hao",
"Chunbo",
""
],
[
"Wang",
"Shuai",
""
],
[
"Yao",
"Jixun",
""
],
[
"Ning",
"Ziqian",
""
],
[
"Meng",
"Meng",
""
],
[
"Luan",
"Jian",
""
],
[
"Xie",
"Lei",
""
]
] |
2507.12901
|
Xiaoke Zhao
|
Xiaoke Zhao, Zhaowen Zhou, Lin Chen, Lihong Wang, Zhiyi Huang, Kaiyuan Zheng, Yanjun Zheng, Xiyang Du, Longfei Liao, Jiawei Liu, Xiang Qi, Bo Zhang, Peng Zhang, Wei Wang, Zhe Li
|
Agentar-DeepFinance-100K: A Large-Scale Financial Dataset via Systematic Chain-of-Thought Synthesis Optimization
| null | null | null | null |
cs.CE
|
http://creativecommons.org/licenses/by/4.0/
|
Recent advancements in large language models (LLMs) have demonstrated remarkable general reasoning capabilities, holding significant potential for applications in the financial domain, a field that requires robust and reliable reasoning. It has been demonstrated that distilling high-quality chain-of-thought (CoT) rationales from advanced general reasoning models offers a promising and efficient path to the financial reasoning model. However, existing CoT synthesis methods suffer from shallow CoT sampling, leaving the question of how to construct a well-designed knowledge space for finance reasoning unexplored. In this paper, we present Agentar-DeepFinance-100K, a large-scale financial reasoning dataset characterized by its systematic CoT synthesis optimization. We first introduce a comprehensive CoT synthesis pipeline featuring Multi-perspective Knowledge Extraction (MKE) and Self-Corrective Rewriting (SCR) to generate exhaustive and deep financial reasoning trajectories. Furthermore, a systematic investigation, termed CoT Cube, is conducted to analyze critical factors that influence CoT effectiveness, such as necessity, length and synthesizer, yielding valuable insights for high-quality financial CoT construction. Experiments demonstrate that models trained on our Agentar-DeepFinance-100K achieve significant improvements on financial benchmarks. We publicly release Agentar-DeepFinance-100K , hoping to advance the research in financial reasoning models.
|
[
{
"version": "v1",
"created": "Thu, 17 Jul 2025 08:40:45 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:50:24 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zhao",
"Xiaoke",
""
],
[
"Zhou",
"Zhaowen",
""
],
[
"Chen",
"Lin",
""
],
[
"Wang",
"Lihong",
""
],
[
"Huang",
"Zhiyi",
""
],
[
"Zheng",
"Kaiyuan",
""
],
[
"Zheng",
"Yanjun",
""
],
[
"Du",
"Xiyang",
""
],
[
"Liao",
"Longfei",
""
],
[
"Liu",
"Jiawei",
""
],
[
"Qi",
"Xiang",
""
],
[
"Zhang",
"Bo",
""
],
[
"Zhang",
"Peng",
""
],
[
"Wang",
"Wei",
""
],
[
"Li",
"Zhe",
""
]
] |
2507.12922
|
Raúl de la Fuente Marcos
|
R. de la Fuente Marcos, J. Licandro, M. R. Alarcon, M Serra-Ricart, J. de Leon, C. de la Fuente Marcos, G. Lombardi, A. Tejero, A. Cabrera-Lavers, S. Guerra Arencibia, and I. Ruiz Cejudo
|
Assessing interstellar comet 3I/ATLAS with the 10.4 m Gran Telescopio Canarias and the Two-meter Twin Telescope
|
7 pages, 7 figures, 2 tables. Abstract abridged as per arXiv directive ("The abstract field cannot be longer than 1,920 characters"). Accepted by A&A Letters
| null |
10.1051/0004-6361/202556439
| null |
astro-ph.EP astro-ph.GA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Context. Theories of the formation and evolution of small bodies in planetary systems predict that they may escape into interstellar space at any time. After having characterized just two such interlopers -1I/2017 U1 (Oumuamua) and 2I/Borisov more questions were raised than answered. Assessing the recently discovered interstellar comet 3I/ATLAS will only broaden our understanding of this complex topic.
Aims. Here, we investigate the spectral, cometary, and rotational properties of 3I/ATLAS as well as its dynamical context.
Methods. We identified the spectral type of 3I/ATLAS from the visible reflectance spectrum and used photometric observations to derive its level of activity and rotational properties. Observational data were obtained with the OSIRIS camera spectrograph at the 10.4 m Gran Telescopio Canarias and the Two-meter Twin Telescope. We used N-body simulations and statistical analyses of Gaia DR3 data to investigate the origin of 3I/ATLAS and its Galactic background.
Results. Interstellar comet 3I/ATLAS has a visible spectrum consistent with that of a D-type asteroid, and has a spectral slope of 14.6%/1000 A in the 5000-9000 A range, which is similar to the one of Oumuamua but redder than that of 2I/Borisov and most solar system comets. It has a conspicuous coma and its rotation period is 16.79 h. The heliocentric components of its Galactic velocity were (U, V, W) = (-51.233, -19.456, 18.930) km/s with a radiant in Sagittarius. The analysis of a sample of kinematic analogs of 3I/ATLAS extracted from Gaia DR3 suggests that its parent system is part of the Galactic thin disk and includes a solar-like star with slightly sub-solar metallicity.
|
[
{
"version": "v1",
"created": "Thu, 17 Jul 2025 09:03:52 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 17:28:07 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Marcos",
"R. de la Fuente",
""
],
[
"Licandro",
"J.",
""
],
[
"Alarcon",
"M. R.",
""
],
[
"Serra-Ricart",
"M",
""
],
[
"de Leon",
"J.",
""
],
[
"Marcos",
"C. de la Fuente",
""
],
[
"Lombardi",
"G.",
""
],
[
"Tejero",
"A.",
""
],
[
"Cabrera-Lavers",
"A.",
""
],
[
"Arencibia",
"S. Guerra",
""
],
[
"Cejudo",
"I. Ruiz",
""
]
] |
2507.12997
|
Sreekantham K. Rithvik
|
S. K. Rithvik
|
A Modular PyTheus Quantum Network Interpreter: Automated Analysis and Visualization of Optimized Quantum Architectures
| null | null | null | null |
quant-ph
|
http://creativecommons.org/licenses/by-sa/4.0/
|
We present a modular interpreter for PyTheus-optimized quantum networks that automatically analyzes and visualizes complex quantum architectures discovered through automated optimization. The interpreter addresses the critical challenge of understanding machine-designed quantum networks by providing robust algorithms for functional role identification, graph-theoretical analysis, and physically meaningful visualization across the major classes of PyTheus-generated networks. Our interpreter accepts both file-based and in-memory network representations, automatically identifies sources, detectors, beam splitters, and ancillas through priority-based classification, and generates coordinated native graph plots and optical table representations. We demonstrate the interpreter's capabilities through two complementary approaches: (1) analysis of a newly developed five-node quantum key distribution network that reveals distributed source architecture and dual-role node functionality, and (2) comprehensive validation using existing PyTheus examples including W4 state generation, heralded Bell state preparation, and GHZ state networks. The interpreter successfully handles complex connectivity patterns across diverse quantum network architectures within the tested classes, avoids visualization artifacts, and provides validation mechanisms for architectural consistency. Our primary contribution is the development of robust modular interpretation algorithms that can analyze the major classes of PyTheus-generated quantum networks, enabling better understanding of automated quantum architecture design.
|
[
{
"version": "v1",
"created": "Thu, 17 Jul 2025 11:13:29 GMT"
},
{
"version": "v2",
"created": "Fri, 18 Jul 2025 08:36:09 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 10:21:55 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Rithvik",
"S. K.",
""
]
] |
2507.13069
|
Harsha K H
|
Harsha K. H., Annapurni Subramaniam, S. R. Dhanush, and Harharan D. S
|
A kinematic and structural study of young open clusters in the Milky Way Galaxy using Gaia DR3 catalogue
|
Accepted for publication in the Journal of Astrophysics and Astronomy (JoAA)
| null | null | null |
astro-ph.SR astro-ph.GA
|
http://creativecommons.org/licenses/by/4.0/
|
We aim to identify the cluster members, estimate cluster properties, study the dynamical state of the clusters as a function of mass, trace the existence of dynamical effects in massive stars, and check for spatial patterns of members in young clusters. We studied 14 young open clusters located within 1 kpc using the data from Gaia DR3 with the membership estimated using the GMM method. The cluster parameters such as age, distance, metallicity, and extinction were estimated by fitting PARSEC isochrones to the CMDs. These clusters are found to have ages between 6-90 Myr, located between 334-910 pc, covering a mass range, of 0.13 to 13.77 solar mass. In five of these clusters, stars from F to M spectral type show increasing velocity dispersion, a signature for dynamical relaxation. We detect high proper motion for B and A-type stars, possible walkaway stars in the other five clusters, Alessi Teutsch 5, ASCC 16, ASCC 21, IC 2395, and NGC 6405. We demonstrate the existence of mass-dependent velocity dispersion in young clusters suggestive of dynamical relaxation. The typical range of transverse velocity dispersion is found to be 0.40 - 0.70 km/s for young clusters.
|
[
{
"version": "v1",
"created": "Thu, 17 Jul 2025 12:36:03 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 14:52:13 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"H.",
"Harsha K.",
""
],
[
"Subramaniam",
"Annapurni",
""
],
[
"Dhanush",
"S. R.",
""
],
[
"S",
"Harharan D.",
""
]
] |
2507.13238
|
Sunny Rai
|
Ashray Gupta and Rohan Joseph and Sunny Rai
|
Multilingual LLMs Are Not Multilingual Thinkers: Evidence from Hindi Analogy Evaluation
| null | null | null | null |
cs.CL cs.AI
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Analogies test a model's ability to infer implicit relationships between concepts, making them a key benchmark for evaluating reasoning capabilities. While large language models (LLMs) are widely evaluated for reasoning in English, their abilities in Indic languages remain understudied, limiting our understanding of whether these models generalize across languages. To address this gap, we introduce a new Hindi Analogy Test Set (HATS), comprising 405 multiple-choice questions sourced from Indian government exams. We benchmark state-of-the-art multilingual LLMs using various prompting strategies and introduce a grounded Chain of Thought approach that leverages cognitive theories of analogical reasoning. This approach improves model performance on Hindi analogy questions. Our experiments show that models perform best with English prompts, irrespective of the prompting strategy. Our test set addresses the lack of a critical resource to evaluate LLM reasoning capabilities in Hindi.
|
[
{
"version": "v1",
"created": "Thu, 17 Jul 2025 15:47:49 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 21:50:22 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Gupta",
"Ashray",
""
],
[
"Joseph",
"Rohan",
""
],
[
"Rai",
"Sunny",
""
]
] |
2507.13921
|
Jenny Frediani
|
Jenny Frediani, Arjan Bik, Mar\'ia Claudia Ram\'irez-Tannus, Rens Waters, Konstantin V. Getman, Eric D. Feigelson, Bayron Portilla-Revelo, Beno\^it Tabone, Thomas J. Haworth, Andrew Winter, Thomas Henning, Giulia Perotti, Alexis Brandeker, Germ\'an Chaparro, Pablo Cuartas-Restrepo, Sebasti\'an Hern\'andez, Michael A. Kuhn, Thomas Preibisch, Veronica Roccatagliata, Sierk E. van Terwisga, and Peter Zeidler
|
XUE 10. The CO2-rich terrestrial planet-forming region of an externally irradiated Herbig disk
|
20 pages, 13 figures, 4 tables
| null | null | null |
astro-ph.EP
|
http://creativecommons.org/licenses/by/4.0/
|
We investigate the James Webb Space Telescope (JWST) MIRI MRS gas molecular content of an externally irradiated Herbig disk, the F-type XUE 10 source, in the context of the eXtreme UV Environments (XUE) program. XUE 10 belongs to the massive star cluster NGC 6357 (1.69 kpc), where it is exposed to an external far-ultraviolet (FUV) radiation $\approx$ 10$^3$ times stronger than in the Solar neighborhood. We modeled the molecular features in the mid-infrared spectrum with Local Thermodynamic Equilibrium (LTE) 0D slab models. We derived basic parameters of the stellar host from a VLT FORS2 optical spectrum using PHOENIX stellar templates. We detect bright CO2 gas with the first simultaneous detection (> 5$\sigma$) of four isotopologues (12CO2, 13CO2, 16O12C18O, 16O12C17O) in a protoplanetary disk. We also detect faint CO emission (2$\sigma$) and the HI Pf$\alpha$ line (8$\sigma$). We also place strict upper limits on the water content, finding a total column density $\lesssim$ 10$^{18}$ cm$^{-2}$. The CO2 species trace low gas temperatures (300-370 K) with a range of column densities of 7.4 $\times$ 10$^{17}$ cm$^{-2}$ (16O12C17O)-1.3 $\times$ 10$^{20}$ cm$^{-2}$ (12CO2) in an equivalent emitting radius of 1.15 au. The emission of 13CO2 is likely affected by line optical depth effects. 16O12C18O and 16O12C17O abundances may be isotopically anomalous compared to the 16O/18O and 16O/17O ratios measured in the interstellar medium and the Solar System. We propose that the mid-infrared spectrum of XUE 10 is explained by H2O removal either via advection or strong photo-dissociation by stellar UV irradiation, and enhanced local CO2 gas-phase production. Outer disk truncation supports the observed CO2-H2O dichotomy. A CO2 vapor enrichment in 18O and 17O can be explained by means of external UV irradiation and early on (10$^{4-5}$ yr) delivery of isotopically anomalous water ice to the inner disk.
|
[
{
"version": "v1",
"created": "Fri, 18 Jul 2025 13:51:31 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:34:39 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Frediani",
"Jenny",
""
],
[
"Bik",
"Arjan",
""
],
[
"Ramírez-Tannus",
"María Claudia",
""
],
[
"Waters",
"Rens",
""
],
[
"Getman",
"Konstantin V.",
""
],
[
"Feigelson",
"Eric D.",
""
],
[
"Portilla-Revelo",
"Bayron",
""
],
[
"Tabone",
"Benoît",
""
],
[
"Haworth",
"Thomas J.",
""
],
[
"Winter",
"Andrew",
""
],
[
"Henning",
"Thomas",
""
],
[
"Perotti",
"Giulia",
""
],
[
"Brandeker",
"Alexis",
""
],
[
"Chaparro",
"Germán",
""
],
[
"Cuartas-Restrepo",
"Pablo",
""
],
[
"Hernández",
"Sebastián",
""
],
[
"Kuhn",
"Michael A.",
""
],
[
"Preibisch",
"Thomas",
""
],
[
"Roccatagliata",
"Veronica",
""
],
[
"van Terwisga",
"Sierk E.",
""
],
[
"Zeidler",
"Peter",
""
]
] |
2507.13991
|
Daniele Viganò
|
Daniele Vigan\`o, Soumya Sengupta, Cl\`audia Soriano-Guerrero, Rosalba Perna, Albert Elias-L\'opez, Sandeep Kumar, Taner Akg\"un
|
Inflated hot Jupiters: Inferring average atmospheric velocity via Ohmic models coupled with internal dynamo evolution
|
Accepted for publication in A&A main journal. 22 pages, 14 figures, 1 table. Comments welcome
| null |
10.1051/0004-6361/202555219
| null |
astro-ph.EP astro-ph.SR
|
http://creativecommons.org/licenses/by/4.0/
|
The inflated radii observed in hundreds of hot Jupiters (HJ) represent a long-standing open issue. In this study, we quantitatively investigate this phenomenon within the framework of Ohmic dissipation arising from magnetic induction in the atmosphere, one of the most promising mechanisms for explaining the radius anomaly. We simulate the evolution of irradiated giant planets with MESA, spanning the observed range of masses and equilibrium temperatures, incorporating an internal source of Ohmic dissipation that extends to deep layers of the envelope. We infer average atmospheric wind intensities, averaged in the region $p < 10$ bar, in the range 0.01-1 km/s in order to reproduce the range of observed radii, decreasing roughly linearly with planetary mass, and much more steeply with equilibrium temperature. This is consistent with the expected effects of magnetic drag from the induced field, which is higher for more intense irradiation, via conductivity, and for larger masses, which have higher dynamo fields. Due to the evolution of the dynamo field and the proportionality of the induced currents on it, the Ohmic efficiency typically decreases by at least one order of magnitude from 0.1 to 10 Gyr, at contrast with the common assumption of a constant-in-time value. Notably, the extent of the main convective region, and the associated heat flux supporting the dynamo, is reduced in the presence of strong Ohmic dissipation, which in turn depends on the dynamo field strength, generating a non-trivial coupling of the latter with the atmospheric induction, potentially leading to an oscillatory behaviour of the field strength. These findings remain generally valid even when accounting for a long-term increase in the main-sequence host star luminosity, although this case can more readily lead to HJ re-inflation, consistent with previous studies.
|
[
{
"version": "v1",
"created": "Fri, 18 Jul 2025 14:58:03 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 13:56:45 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Viganò",
"Daniele",
""
],
[
"Sengupta",
"Soumya",
""
],
[
"Soriano-Guerrero",
"Clàudia",
""
],
[
"Perna",
"Rosalba",
""
],
[
"Elias-López",
"Albert",
""
],
[
"Kumar",
"Sandeep",
""
],
[
"Akgün",
"Taner",
""
]
] |
2507.14009
|
Francesco Fournier-Facio
|
Francesco Fournier-Facio
|
Stable commutator length on free $\mathbb{Q}$-groups
|
13 pages. v2: removed Remark 3.4, which had a mistake, and added Corollary 4.4
| null | null | null |
math.GR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We study stable commutator length on free $\mathbb{Q}$-groups. We prove that every non-identity element has positive stable commutator length, and that the corresponding free group embeds isometrically. We deduce that a non-abelian free $\mathbb{Q}$-group has an infinite-dimensional space of homogeneous quasimorphisms modulo homomorphisms, answering a question of Casals-Ruiz, Garreta, and de la Nuez Gonz{\'a}lez. We conjecture that stable commutator length is rational on free $\mathbb{Q}$-groups. This is connected to the long-standing problem of rationality on surface groups: indeed, we show that free $\mathbb{Q}$-groups contain isometrically embedded copies of non-orientable surface groups.
|
[
{
"version": "v1",
"created": "Fri, 18 Jul 2025 15:20:58 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 07:39:59 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Fournier-Facio",
"Francesco",
""
]
] |
2507.14075
|
Paul R. Berman
|
Paul R. Berman and Peter W. Milonni
|
Spontaneous emission in dipole approximation -- revisited
|
Found a possible error in using Eq. (18) to calculate source field expressions in Sec. IV and VI
| null | null | null |
quant-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Spontaneous emission in dipole approximation is studied theoretically using both source-field theory and a Schrodinger picture approach. Using source-field theory we obtain formal equations for the Poynting vector and energy density without making the rotating wave approximation (RWA) and Weisskopf-Wigner approximation (WWA). The initial condition at t=0 is one in which the atom is in an excited state and the field in the vacuum state. The source-field expressions are evaluated within the the RWA and WWA and are found to satisfy Poynting's theorem. To explore the consequences of not making the RWA and WWA, the Poynting vector and energy density are calculated using perturbation theory. We use a Schrodinger picture approach and essentially reproduce and complement the results of Compagno, Passante, and Persico [J. Mod. Optics 37:8, 1377 (2007)] and those of Power and Thirunamachandran [Phys. Rev. A 45, 54 (1992)] obtained using a Heisenberg picture approach. The theory involves a sum over field mode frequencies and both finite cutoffs and convergence factors are used to carry out the sums. It is shown that the perturbation theory calculation leads to unphysical values for atomic state populations for all times when a sum over all field frequencies is taken, even if a convergence factor is used. It is also proved that the fields calculated using source-field theory always satisfy Poynting's theorem for ct not equal to R, where R is the distance from the atom.
|
[
{
"version": "v1",
"created": "Fri, 18 Jul 2025 16:52:57 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 21:24:57 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Berman",
"Paul R.",
""
],
[
"Milonni",
"Peter W.",
""
]
] |
2507.14314
|
Dominik Ã
 ipek
|
Marija An{\dj}eli\'c, Dominik \v{S}ipek, Laura Majer, Jan \v{S}najder
|
What Makes You CLIC: Detection of Croatian Clickbait Headlines
|
Accepted at Slavic NLP 2025
| null | null | null |
cs.CL
|
http://creativecommons.org/licenses/by-sa/4.0/
|
Online news outlets operate predominantly on an advertising-based revenue model, compelling journalists to create headlines that are often scandalous, intriguing, and provocative -- commonly referred to as clickbait. Automatic detection of clickbait headlines is essential for preserving information quality and reader trust in digital media and requires both contextual understanding and world knowledge. For this task, particularly in less-resourced languages, it remains unclear whether fine-tuned methods or in-context learning (ICL) yield better results. In this paper, we compile CLIC, a novel dataset for clickbait detection of Croatian news headlines spanning a 20-year period and encompassing mainstream and fringe outlets. We fine-tune the BERTi\'c model on this task and compare its performance to LLM-based ICL methods with prompts both in Croatian and English. Finally, we analyze the linguistic properties of clickbait. We find that nearly half of the analyzed headlines contain clickbait, and that finetuned models deliver better results than general LLMs.
|
[
{
"version": "v1",
"created": "Fri, 18 Jul 2025 18:39:07 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 17:10:17 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Anđelić",
"Marija",
""
],
[
"Šipek",
"Dominik",
""
],
[
"Majer",
"Laura",
""
],
[
"Šnajder",
"Jan",
""
]
] |
2507.14349
|
Anthony Badea
|
Anthony Badea, Austin Baty, Hannah Bossi, Yu-Chen Chen, Yi Chen, Jingyu Zhang, Gian Michele Innocenti, Marcello Maggi, Chris McGinn, Michael Peters, Tzu-An Sheng, Vinicius Mikuni, Matthew Avaylon, Patrick Komiske, Eric Metodiev, Jesse Thaler, Benjamin Nachman, Yen-Jie Lee
|
Analysis note: measurement of thrust in $e^{+}e^{-}$ collisions at $\sqrt{s}$ = 91 GeV with archived ALEPH data
| null | null | null | null |
hep-ex
|
http://creativecommons.org/licenses/by/4.0/
|
A measurement of the thrust distribution in $e^{+}e^{-}$ collisions at $\sqrt{s} = 91.2$ GeV with archived data from the ALEPH experiment at the Large Electron-Positron Collider is presented. The thrust distribution is reconstructed from charged and neutral particles resulting from hadronic $Z$-boson decays. For the first time with $e^{+}e^{-}$ data, detector effects are corrected using a machine learning based method for unbinned unfolding. The measurement provides new input for resolving current discrepancies between theoretical calculations and experimental determinations of $\alpha_{s}$, constraining non-perturbative effects through logarithmic moments, developing differential hadronization models, and enabling new precision studies using the archived data.
|
[
{
"version": "v1",
"created": "Fri, 18 Jul 2025 19:56:14 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 18:55:58 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Badea",
"Anthony",
""
],
[
"Baty",
"Austin",
""
],
[
"Bossi",
"Hannah",
""
],
[
"Chen",
"Yu-Chen",
""
],
[
"Chen",
"Yi",
""
],
[
"Zhang",
"Jingyu",
""
],
[
"Innocenti",
"Gian Michele",
""
],
[
"Maggi",
"Marcello",
""
],
[
"McGinn",
"Chris",
""
],
[
"Peters",
"Michael",
""
],
[
"Sheng",
"Tzu-An",
""
],
[
"Mikuni",
"Vinicius",
""
],
[
"Avaylon",
"Matthew",
""
],
[
"Komiske",
"Patrick",
""
],
[
"Metodiev",
"Eric",
""
],
[
"Thaler",
"Jesse",
""
],
[
"Nachman",
"Benjamin",
""
],
[
"Lee",
"Yen-Jie",
""
]
] |
2507.14424
|
Richard F. Bartels
|
Richard F. Bartels
|
Cohen-Macaulay approximations and the $\text{SC}_r$-condition
|
24 pages. Clarified definition of FID hull. Improved results in section 2
| null | null | null |
math.AC
|
http://creativecommons.org/licenses/by/4.0/
|
We study the relation between MCM approximations and FID hulls of modules over a Cohen-Macaulay local ring $R$ with canonical module, specifically when $R$ is generically Gorenstein. We then generalize a result of Kato, who proved that a Gorenstein complete local ring $R$ satisfies the $\text{SC}_{2}$-condition if and only if $R$ is a UFD. For $r \geq 3$, we prove a criterion for when an MCM $R$-module $M$ satisfies the $\text{SC}_{r}$-condition, assuming that its first syzygy $\Omega_{R}^{1}(M)$ satisfies the $\text{SC}_{r-1}$-condition.
|
[
{
"version": "v1",
"created": "Sat, 19 Jul 2025 00:52:51 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 16:34:58 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Bartels",
"Richard F.",
""
]
] |
2507.14516
|
Jeyoung Lee
|
Jeyoung Lee and Hochul Kang
|
SDSC:A Structure-Aware Metric for Semantic Signal Representation Learning
| null | null | null | null |
cs.LG cs.AI cs.LO
|
http://creativecommons.org/licenses/by/4.0/
|
We propose the Signal Dice Similarity Coefficient (SDSC), a structure-aware metric function for time series self-supervised representation learning. Most Self-Supervised Learning (SSL) methods for signals commonly adopt distance-based objectives such as mean squared error (MSE), which are sensitive to amplitude, invariant to waveform polarity, and unbounded in scale. These properties hinder semantic alignment and reduce interpretability. SDSC addresses this by quantifying structural agreement between temporal signals based on the intersection of signed amplitudes, derived from the Dice Similarity Coefficient (DSC).Although SDSC is defined as a structure-aware metric, it can be used as a loss by subtracting from 1 and applying a differentiable approximation of the Heaviside function for gradient-based optimization. A hybrid loss formulation is also proposed to combine SDSC with MSE, improving stability and preserving amplitude where necessary. Experiments on forecasting and classification benchmarks demonstrate that SDSC-based pre-training achieves comparable or improved performance over MSE, particularly in in-domain and low-resource scenarios. The results suggest that structural fidelity in signal representations enhances the semantic representation quality, supporting the consideration of structure-aware metrics as viable alternatives to conventional distance-based methods.
|
[
{
"version": "v1",
"created": "Sat, 19 Jul 2025 07:32:00 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 07:48:25 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Lee",
"Jeyoung",
""
],
[
"Kang",
"Hochul",
""
]
] |
2507.14562
|
Shuai Wang
|
Yuanling Niu, Shuai Wang, Ying Zhang
|
1/2 order convergence rate of Euler-type methods for time-changed stochastic differential equations with super-linearly growing drift and diffusion coefficients
| null | null | null | null |
math.NA cs.NA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This paper investigates the convergence rates of two Euler-type methods for a class of time-changed stochastic differential equations with super-linearly growing drift and diffusion coefficients. Building upon existing research, we adapt the backward Euler method to time-changed stochastic differential equations where both coefficients exhibit super-linear growth and introduce an explicit counterpart, the projected Euler method. It is shown that both methods achieve the optimal strong convergence rate of order 1/2 in the mean-square sense for this class of equations. Numerical simulations confirm the theoretical findings
|
[
{
"version": "v1",
"created": "Sat, 19 Jul 2025 09:56:04 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 01:34:37 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Niu",
"Yuanling",
""
],
[
"Wang",
"Shuai",
""
],
[
"Zhang",
"Ying",
""
]
] |
2507.14660
|
Qibing Ren
|
Qibing Ren, Sitao Xie, Longxuan Wei, Zhenfei Yin, Junchi Yan, Lizhuang Ma, Jing Shao
|
When Autonomy Goes Rogue: Preparing for Risks of Multi-Agent Collusion in Social Systems
|
Code is available at https://github.com/renqibing/MultiAgent4Collusion
| null | null | null |
cs.AI cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Recent large-scale events like election fraud and financial scams have shown how harmful coordinated efforts by human groups can be. With the rise of autonomous AI systems, there is growing concern that AI-driven groups could also cause similar harm. While most AI safety research focuses on individual AI systems, the risks posed by multi-agent systems (MAS) in complex real-world situations are still underexplored. In this paper, we introduce a proof-of-concept to simulate the risks of malicious MAS collusion, using a flexible framework that supports both centralized and decentralized coordination structures. We apply this framework to two high-risk fields: misinformation spread and e-commerce fraud. Our findings show that decentralized systems are more effective at carrying out malicious actions than centralized ones. The increased autonomy of decentralized systems allows them to adapt their strategies and cause more damage. Even when traditional interventions, like content flagging, are applied, decentralized groups can adjust their tactics to avoid detection. We present key insights into how these malicious groups operate and the need for better detection systems and countermeasures. Code is available at https://github.com/renqibing/RogueAgent.
|
[
{
"version": "v1",
"created": "Sat, 19 Jul 2025 15:17:30 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 06:00:02 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Ren",
"Qibing",
""
],
[
"Xie",
"Sitao",
""
],
[
"Wei",
"Longxuan",
""
],
[
"Yin",
"Zhenfei",
""
],
[
"Yan",
"Junchi",
""
],
[
"Ma",
"Lizhuang",
""
],
[
"Shao",
"Jing",
""
]
] |
2507.14672
|
Hervé Zwirn
|
Herve Zwirn
|
Are Events Absolute?
| null | null | null | null |
quant-ph physics.hist-ph
|
http://creativecommons.org/licenses/by/4.0/
|
The Wigner's Friend thought experiment stands as one of the most intellectually provocative and challenging conceptual puzzles in quantum mechanics. It compels us to confront profound questions concerning the fundamental nature of reality, the very act of observation, and the possible role that consciousness might play within the quantum measurement process. This article gives a general presentation, beginning with Eugene Wigner's seminal proposal of the original thought experiment. In this paper, we explore its initial implications, which shook the foundations of classical physics, and then progress to an examination of the recent theoretical advancements and the ingenious extended versions of the experiment. The recent versions seem to imply that it is no more possible to consider events as absolute.
|
[
{
"version": "v1",
"created": "Sat, 19 Jul 2025 15:56:40 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 09:11:36 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zwirn",
"Herve",
""
]
] |
2507.14679
|
Zhijie Wang
|
Zhijie Wang, Zixin Xu, Zhiyuan Pan
|
GCC-Spam: Spam Detection via GAN, Contrastive Learning, and Character Similarity Networks
| null | null | null | null |
cs.LG cs.AI cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The exponential growth of spam text on the Internet necessitates robust detection mechanisms to mitigate risks such as information leakage and social instability. This work addresses two principal challenges: adversarial strategies employed by spammers and the scarcity of labeled data. We propose a novel spam-text detection framework GCC-Spam, which integrates three core innovations. First, a character similarity network captures orthographic and phonetic features to counter character-obfuscation attacks and furthermore produces sentence embeddings for downstream classification. Second, contrastive learning enhances discriminability by optimizing the latent-space distance between spam and normal texts. Third, a Generative Adversarial Network (GAN) generates realistic pseudo-spam samples to alleviate data scarcity while improving model robustness and classification accuracy. Extensive experiments on real-world datasets demonstrate that our model outperforms baseline approaches, achieving higher detection rates with significantly fewer labeled examples.
|
[
{
"version": "v1",
"created": "Sat, 19 Jul 2025 16:09:48 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 15:46:28 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Wang",
"Zhijie",
""
],
[
"Xu",
"Zixin",
""
],
[
"Pan",
"Zhiyuan",
""
]
] |
2507.14783
|
Jiaming Zhou
|
Derek Li and Jiaming Zhou, Amirreza Kazemi, Qianyi Sun, Abbas Ghaddar, Mohammad Ali Alomrani, Liheng Ma, Yu Luo, Dong Li, Feng Wen, Jianye Hao, Mark Coates, Yingxue Zhang
|
Omni-Thinker: Scaling Cross-Domain Generalization in LLMs via Multi-Task RL with Hybrid Rewards
| null | null | null | null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The advancement of general-purpose artificial intelligence relies on large language models (LLMs) that excel across a wide range of tasks, from structured reasoning to creative generation. However, post-training methods like Supervised Fine-Tuning (SFT) often struggle with generalization, favoring memorization over transferable learning. In this work, we introduce Omni-Thinker, a unified reinforcement learning (RL) framework that enhances LLM performance across diverse tasks by combining rule-based verifiable rewards with generative preference signals via LLM-as-a-Judge evaluations. Our approach enables consistent optimization across task types and scales RL-based training to subjective domains. We further investigate training strategies, demonstrating that a curriculum-based progression that orders tasks from structured to open-ended improves performance and reduces forgetting. Experimental results across four domains reveal that curriculum learning improves performance by 5.2% over joint training and 9.1% over model merging. These results highlight the importance of task-aware sampling and hybrid supervision in scaling RL-based post-training for general-purpose LLMs.
|
[
{
"version": "v1",
"created": "Sun, 20 Jul 2025 01:50:16 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 16:25:54 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Li",
"Derek",
""
],
[
"Zhou",
"Jiaming",
""
],
[
"Kazemi",
"Amirreza",
""
],
[
"Sun",
"Qianyi",
""
],
[
"Ghaddar",
"Abbas",
""
],
[
"Alomrani",
"Mohammad Ali",
""
],
[
"Ma",
"Liheng",
""
],
[
"Luo",
"Yu",
""
],
[
"Li",
"Dong",
""
],
[
"Wen",
"Feng",
""
],
[
"Hao",
"Jianye",
""
],
[
"Coates",
"Mark",
""
],
[
"Zhang",
"Yingxue",
""
]
] |
2507.14905
|
Abubakar Fadul
|
Abubakar M. A. Fadul, Kamber R. Schwarz, Tushar Suhasaria, Jenny K. Calahan, Jane Huang, and Merel L. R. van 't Hoff
|
A deep Search for Ethylene Glycol and Glycolonitrile in the V883 Ori Protoplanetary Disk
|
15 pages, 8 figures, accepted for publication in ApJL
| null |
10.3847/2041-8213/adec6e
| null |
astro-ph.SR astro-ph.EP astro-ph.GA
|
http://creativecommons.org/licenses/by/4.0/
|
Ethylene glycol ($\mathrm{(CH_2OH)_2}$, hereafter EG) and Glycolonitrile ($\mathrm{HOCH_2CN}$, hereafter GN) are considered molecular precursors of nucleic acids. EG is a sugar alcohol and the reduced form of Glycolaldehyde ($\mathrm{CH_2(OH)CHO}$, hereafter GA). GN is considered a key precursor of adenine formation (nucleotide) and can be a precursor of glycine (amino acid). Detections of such prebiotic molecules in the interstellar medium are increasingly common. How much of this complexity endures to the planet formation stage, and thus is already present when planets form, remains largely unknown. Here we report Atacama Large Millimeter/sub-millimeter Array (ALMA) observations in which we tentatively detect EG and GN in the protoplanetary disk around the outbursting protostar V883 Ori. The observed EG emission is best reproduced by a column density of $\mathrm{3.63^{+0.11}_{-0.12} \times 10^{16} \; cm^{-2}}$ and a temperature of at least 300 K. The observed GN emission is best reproduced by a column density of $\mathrm{3.37^{+0.09}_{-0.09} \times 10^{16} \; cm^{-2}}$ and a temperature of $88^{+1.2}_{-1.2}$ K. Comparing the abundance of EG and GN relative to methanol in V883 Ori with other objects, V883 Ori falls between hot cores and comets in terms of increasing complexity. This suggests that the build up of prebiotic molecules continues past the hot core phase into the epoch of planet formation. Nascent planets in such environments may inherit essential building blocks for life, enhancing their potential habitability. Further observations of this protoplanetary disk at higher spectral resolution are required to resolve blended lines and to confirm these tentative detection.
|
[
{
"version": "v1",
"created": "Sun, 20 Jul 2025 10:32:25 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Fadul",
"Abubakar M. A.",
""
],
[
"Schwarz",
"Kamber R.",
""
],
[
"Suhasaria",
"Tushar",
""
],
[
"Calahan",
"Jenny K.",
""
],
[
"Huang",
"Jane",
""
],
[
"Hoff",
"Merel L. R. van 't",
""
]
] |
2507.14937
|
Hugh Kennedy Dr.
|
Hugh L Kennedy
|
Phase-optimised linearly-constrained minimum-variance beamformers
|
Fixed a few minor things spotted since the initial upload
| null | null | null |
eess.SP
|
http://creativecommons.org/licenses/by/4.0/
|
A novel procedure for the determination of the optimal group-delay of a Linearly-Constrained Minimum-Variance (LCMV) beamformer is proposed. Two ways of selecting the optimal delay are recommended: the first is the solution that minimizes the noise power; the second is the solution that minimizes the processing delay. The potential of this hitherto unexplored degree of design freedom is explored using simulated Very-High-Frequency (VHF) communication, and Ultra-High-Frequency (UHF) bistatic radar, applications.
|
[
{
"version": "v1",
"created": "Sun, 20 Jul 2025 12:21:27 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 13:02:46 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Kennedy",
"Hugh L",
""
]
] |
2507.15077
|
Pankaj Bhagwat
|
Pankaj Bhagwat and Eric Marchand
|
Unbiased estimation in one-parameter exponential families for the inverse of the natural parameter with extensions
| null | null | null | null |
math.ST stat.TH
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
For one-parameter continuous exponential families, we identify an unbiased estimator of the inverse of the natural parameter $\theta$ for cases where $\theta > 0$, extending an earlier result of \cite{voinov1985unbiased} applicable to a normal model. We provide various applications for Gamma models, Inverse Gaussian models, distributions obtained by truncation, and ratios of normal means. Moreover, we extend the findings to estimating negative powers $\theta^{-k}$, and more generally to complete monotone functions $q(\theta)$.
|
[
{
"version": "v1",
"created": "Sun, 20 Jul 2025 18:23:54 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 13:48:06 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Bhagwat",
"Pankaj",
""
],
[
"Marchand",
"Eric",
""
]
] |
2507.15103
|
Liet Vo
|
Liet Vo
|
Analysis of fully discrete Crank-Nicolson finite element methods for a stochastic Keller-Segel chemotaxis system with gradient-type multiplicative noise
| null | null | null | null |
math.NA cs.NA
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
We develop and analyze numerical methods for a stochastic Keller-Segel system perturbed by Stratonovich noise, which models chemotactic behavior under randomly fluctuating environmental conditions. The proposed fully discrete scheme couples a Crank-Nicolson time discretization with a splitting mixed finite element method in space. We rigorously prove the stability of the numerical scheme and establish strong convergence rates of order $O(k^{1/2} + k^{-1/2}h^2)$, where $k$ and $h$ denote the time and spatial step sizes, respectively. Notably, the presence of stochastic forcing leads to an inverse dependence on $k$ in the error estimates, distinguishing the convergence behavior from that of the deterministic case. Numerical experiments are presented to validate the theoretical results and demonstrate the effectiveness and accuracy of the proposed methods.
|
[
{
"version": "v1",
"created": "Sun, 20 Jul 2025 19:54:43 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 03:25:52 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Vo",
"Liet",
""
]
] |
2507.15205
|
Xinran Li
|
Xinran Li, Xiujuan Xu, Jiaqi Qiao
|
Long-Short Distance Graph Neural Networks and Improved Curriculum Learning for Emotion Recognition in Conversation
|
Accepted by the 28th European Conference on Artificial Intelligence (ECAI 2025)
| null | null | null |
cs.LG cs.AI cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
Emotion Recognition in Conversation (ERC) is a practical and challenging task. This paper proposes a novel multimodal approach, the Long-Short Distance Graph Neural Network (LSDGNN). Based on the Directed Acyclic Graph (DAG), it constructs a long-distance graph neural network and a short-distance graph neural network to obtain multimodal features of distant and nearby utterances, respectively. To ensure that long- and short-distance features are as distinct as possible in representation while enabling mutual influence between the two modules, we employ a Differential Regularizer and incorporate a BiAffine Module to facilitate feature interaction. In addition, we propose an Improved Curriculum Learning (ICL) to address the challenge of data imbalance. By computing the similarity between different emotions to emphasize the shifts in similar emotions, we design a "weighted emotional shift" metric and develop a difficulty measurer, enabling a training process that prioritizes learning easy samples before harder ones. Experimental results on the IEMOCAP and MELD datasets demonstrate that our model outperforms existing benchmarks.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 03:12:54 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 05:15:18 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Li",
"Xinran",
""
],
[
"Xu",
"Xiujuan",
""
],
[
"Qiao",
"Jiaqi",
""
]
] |
2507.15292
|
Rulin Zhou
|
An Wang, Rulin Zhou, Mengya Xu, Yiru Ye, Longfei Gou, Yiting Chang, Hao Chen, Chwee Ming Lim, Jiankun Wang, Hongliang Ren
|
EndoControlMag: Robust Endoscopic Vascular Motion Magnification with Periodic Reference Resetting and Hierarchical Tissue-aware Dual-Mask Control
| null | null | null | null |
eess.IV cs.AI cs.CV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Visualizing subtle vascular motions in endoscopic surgery is crucial for surgical precision and decision-making, yet remains challenging due to the complex and dynamic nature of surgical scenes. To address this, we introduce EndoControlMag, a training-free, Lagrangian-based framework with mask-conditioned vascular motion magnification tailored to endoscopic environments. Our approach features two key modules: a Periodic Reference Resetting (PRR) scheme that divides videos into short overlapping clips with dynamically updated reference frames to prevent error accumulation while maintaining temporal coherence, and a Hierarchical Tissue-aware Magnification (HTM) framework with dual-mode mask dilation. HTM first tracks vessel cores using a pretrained visual tracking model to maintain accurate localization despite occlusions and view changes. It then applies one of two adaptive softening strategies to surrounding tissues: motion-based softening that modulates magnification strength proportional to observed tissue displacement, or distance-based exponential decay that simulates biomechanical force attenuation. This dual-mode approach accommodates diverse surgical scenarios-motion-based softening excels with complex tissue deformations while distance-based softening provides stability during unreliable optical flow conditions. We evaluate EndoControlMag on our EndoVMM24 dataset spanning four different surgery types and various challenging scenarios, including occlusions, instrument disturbance, view changes, and vessel deformations. Quantitative metrics, visual assessments, and expert surgeon evaluations demonstrate that EndoControlMag significantly outperforms existing methods in both magnification accuracy and visual quality while maintaining robustness across challenging surgical conditions. The code, dataset, and video results are available at https://szupc.github.io/EndoControlMag/.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 06:47:44 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Jul 2025 05:11:23 GMT"
},
{
"version": "v3",
"created": "Wed, 23 Jul 2025 12:04:57 GMT"
},
{
"version": "v4",
"created": "Thu, 24 Jul 2025 13:26:19 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Wang",
"An",
""
],
[
"Zhou",
"Rulin",
""
],
[
"Xu",
"Mengya",
""
],
[
"Ye",
"Yiru",
""
],
[
"Gou",
"Longfei",
""
],
[
"Chang",
"Yiting",
""
],
[
"Chen",
"Hao",
""
],
[
"Lim",
"Chwee Ming",
""
],
[
"Wang",
"Jiankun",
""
],
[
"Ren",
"Hongliang",
""
]
] |
2507.15401
|
Huiyu Zhai
|
Huiyu Zhai, Xingxing Yang, Yalan Ye, Chenyang Li, Bin Fan, Changze Li
|
Rethinking Occlusion in FER: A Semantic-Aware Perspective and Go Beyond
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Facial expression recognition (FER) is a challenging task due to pervasive occlusion and dataset biases. Especially when facial information is partially occluded, existing FER models struggle to extract effective facial features, leading to inaccurate classifications. In response, we present ORSANet, which introduces the following three key contributions: First, we introduce auxiliary multi-modal semantic guidance to disambiguate facial occlusion and learn high-level semantic knowledge, which is two-fold: 1) we introduce semantic segmentation maps as dense semantics prior to generate semantics-enhanced facial representations; 2) we introduce facial landmarks as sparse geometric prior to mitigate intrinsic noises in FER, such as identity and gender biases. Second, to facilitate the effective incorporation of these two multi-modal priors, we customize a Multi-scale Cross-interaction Module (MCM) to adaptively fuse the landmark feature and semantics-enhanced representations within different scales. Third, we design a Dynamic Adversarial Repulsion Enhancement Loss (DARELoss) that dynamically adjusts the margins of ambiguous classes, further enhancing the model's ability to distinguish similar expressions. We further construct the first occlusion-oriented FER dataset to facilitate specialized robustness analysis on various real-world occlusion conditions, dubbed Occlu-FER. Extensive experiments on both public benchmarks and Occlu-FER demonstrate that our proposed ORSANet achieves SOTA recognition performance. Code is publicly available at https://github.com/Wenyuzhy/ORSANet-master.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 09:04:29 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 06:58:54 GMT"
},
{
"version": "v3",
"created": "Thu, 24 Jul 2025 08:24:01 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zhai",
"Huiyu",
""
],
[
"Yang",
"Xingxing",
""
],
[
"Ye",
"Yalan",
""
],
[
"Li",
"Chenyang",
""
],
[
"Fan",
"Bin",
""
],
[
"Li",
"Changze",
""
]
] |
2507.15413
|
Francesc Ballester
|
Francesc Ballester, Ion Errea, Maia G. Vergniory
|
Simphony: A full tight-binding package for lattice vibrations and topological phonon analysis
|
34 pages, 5 figures, source code available at https://github.com/fballestermacia/simphony
| null | null | null |
cond-mat.mtrl-sci
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Simphony is an open-source software package designed for the topological analysis of lattice vibrations based on Wannier tight-binding models. Its primary function is to classify the topology of novel materials by computing bulk and slab phonon band structures, extracting phonon surface spectra, and providing analysis tools such as Wilson loop calculations and Weyl node detection. The workflow is analogous to that of established electronic topology codes like Wannier90 and WannierTools. It also incorporates long-range polar interactions during the wannierization process, making Simphony one of the first tools capable of diagnosing topology in polar insulators.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 09:11:13 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 08:56:01 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Ballester",
"Francesc",
""
],
[
"Errea",
"Ion",
""
],
[
"Vergniory",
"Maia G.",
""
]
] |
2507.15465
|
Jung Ho Ahn
|
Sungmin Yun and Seonyong Park and Hwayong Nam and Younjoo Lee and Gunjun Lee and Kwanhee Kyung and Sangpyo Kim and Nam Sung Kim and Jongmin Kim and Hyungyo Kim and Juhwan Cho and Seungmin Baek and Jung Ho Ahn
|
The New LLM Bottleneck: A Systems Perspective on Latent Attention and Mixture-of-Experts
|
15 pages, 11 figures
| null | null | null |
cs.AR cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Computational workloads composing traditional Transformer models are starkly bifurcated. Multi-Head Attention (MHA) is memory-bound, with low arithmetic intensity, while feedforward layers are compute-bound. This dichotomy has long motivated research into specialized hardware to mitigate the MHA bottleneck.
This paper argues that recent architectural shifts, namely Multi-head Latent Attention (MLA) and Mixture-of-Experts (MoE), challenge the premise of specialized attention hardware. We make two key observations. First, the arithmetic intensity of MLA is over two orders of magnitude greater than that of MHA, shifting it close to a compute-bound regime well-suited for modern accelerators like GPUs. Second, by distributing MoE experts across a pool of accelerators, their arithmetic intensity can be tuned through batching to match that of the dense layers, creating a more balanced computational profile.
These findings reveal a diminishing need for specialized attention hardware. The central challenge for next-generation Transformers is no longer accelerating a single memory-bound layer. Instead, the focus must shift to designing balanced systems with sufficient compute, memory capacity, memory bandwidth, and high-bandwidth interconnects to manage the diverse demands of large-scale models.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 10:18:33 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 20:55:41 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Yun",
"Sungmin",
""
],
[
"Park",
"Seonyong",
""
],
[
"Nam",
"Hwayong",
""
],
[
"Lee",
"Younjoo",
""
],
[
"Lee",
"Gunjun",
""
],
[
"Kyung",
"Kwanhee",
""
],
[
"Kim",
"Sangpyo",
""
],
[
"Kim",
"Nam Sung",
""
],
[
"Kim",
"Jongmin",
""
],
[
"Kim",
"Hyungyo",
""
],
[
"Cho",
"Juhwan",
""
],
[
"Baek",
"Seungmin",
""
],
[
"Ahn",
"Jung Ho",
""
]
] |
2507.15504
|
Bingqing Zhang
|
Bingqing Zhang, Zhuo Cao, Heming Du, Yang Li, Xue Li, Jiajun Liu, Sen Wang
|
Quantifying and Narrowing the Unknown: Interactive Text-to-Video Retrieval via Uncertainty Minimization
|
Accepted by ICCV 2025
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Despite recent advances, Text-to-video retrieval (TVR) is still hindered by multiple inherent uncertainties, such as ambiguous textual queries, indistinct text-video mappings, and low-quality video frames. Although interactive systems have emerged to address these challenges by refining user intent through clarifying questions, current methods typically rely on heuristic or ad-hoc strategies without explicitly quantifying these uncertainties, limiting their effectiveness. Motivated by this gap, we propose UMIVR, an Uncertainty-Minimizing Interactive Text-to-Video Retrieval framework that explicitly quantifies three critical uncertainties-text ambiguity, mapping uncertainty, and frame uncertainty-via principled, training-free metrics: semantic entropy-based Text Ambiguity Score (TAS), Jensen-Shannon divergence-based Mapping Uncertainty Score (MUS), and a Temporal Quality-based Frame Sampler (TQFS). By adaptively generating targeted clarifying questions guided by these uncertainty measures, UMIVR iteratively refines user queries, significantly reducing retrieval ambiguity. Extensive experiments on multiple benchmarks validate UMIVR's effectiveness, achieving notable gains in Recall@1 (69.2\% after 10 interactive rounds) on the MSR-VTT-1k dataset, thereby establishing an uncertainty-minimizing foundation for interactive TVR.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 11:12:39 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 10:32:09 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zhang",
"Bingqing",
""
],
[
"Cao",
"Zhuo",
""
],
[
"Du",
"Heming",
""
],
[
"Li",
"Yang",
""
],
[
"Li",
"Xue",
""
],
[
"Liu",
"Jiajun",
""
],
[
"Wang",
"Sen",
""
]
] |
2507.15511
|
Jesus Salas
|
Jesus Salas
|
Certificate-Sensitive Subset Sum: Realizing Instance Complexity
|
14 pages + appendix. Companion to arXiv:2503.20162 ("Beyond Worst-Case Subset Sum: An Adaptive, Structure-Aware Solver with Sub-2^{n/2} Enumeration"
| null | null | null |
cs.CC cs.DS
|
http://creativecommons.org/licenses/by/4.0/
|
We present, to our knowledge, the first deterministic, certificate-sensitive algorithm for a canonical NP-complete problem whose runtime provably adapts to the structure of each input. For a Subset-Sum instance $(S, t)$, let $\Sigma(S)$ denote the set of distinct subset sums and define $U = |\Sigma(S)|$. This set serves as an information-theoretically minimal witness, the instance-complexity (IC) certificate.
Our solver, IC-SubsetSum, enumerates every element of $\Sigma(S)$ in deterministic time $O(U \cdot n^2)$ and space $O(U \cdot n)$. A randomized variant achieves expected runtime $O(U \cdot n)$. The algorithm's complexity is thus directly governed by the certificate size, and this structure-sensitive performance is paired with a guaranteed worst-case runtime of $O^*(2^{n/2 - \varepsilon})$ for some constant $\varepsilon > 0$, the first such result to strictly outperform classical methods on every instance.
We revisit fine-grained reductions that rely on the classical $2^{n/2}$ hardness of SubsetSum and show that these arguments hold only for collision-free instances where $U$ is maximal. IC-SubsetSum reframes this barrier structurally and introduces a new paradigm for certificate-sensitive algorithms across NP-complete problems.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 11:26:38 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 14:54:46 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Salas",
"Jesus",
""
]
] |
2507.15541
|
Jongmin Shin
|
Jongmin Shin, Enki Cho, Ka Young Kim, Jung Yong Kim, Seong Tae Kim, and Namkee Oh
|
Towards Holistic Surgical Scene Graph
|
Accepted to MICCAI 2025
| null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Surgical scene understanding is crucial for computer-assisted intervention systems, requiring visual comprehension of surgical scenes that involves diverse elements such as surgical tools, anatomical structures, and their interactions. To effectively represent the complex information in surgical scenes, graph-based approaches have been explored to structurally model surgical entities and their relationships. Previous surgical scene graph studies have demonstrated the feasibility of representing surgical scenes using graphs. However, certain aspects of surgical scenes-such as diverse combinations of tool-action-target and the identity of the hand operating the tool-remain underexplored in graph-based representations, despite their importance. To incorporate these aspects into graph representations, we propose Endoscapes-SG201 dataset, which includes annotations for tool-action-target combinations and hand identity. We also introduce SSG-Com, a graph-based method designed to learn and represent these critical elements. Through experiments on downstream tasks such as critical view of safety assessment and action triplet recognition, we demonstrated the importance of integrating these essential scene graph components, highlighting their significant contribution to surgical scene understanding. The code and dataset are available at https://github.com/ailab-kyunghee/SSG-Com
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 12:10:42 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 00:51:53 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Shin",
"Jongmin",
""
],
[
"Cho",
"Enki",
""
],
[
"Kim",
"Ka Young",
""
],
[
"Kim",
"Jung Yong",
""
],
[
"Kim",
"Seong Tae",
""
],
[
"Oh",
"Namkee",
""
]
] |
2507.15551
|
Yuchen Jiang
|
Jie Zhu, Zhifang Fan, Xiaoxie Zhu, Yuchen Jiang, Hangyu Wang, Xintian Han, Haoran Ding, Xinmin Wang, Wenlin Zhao, Zhen Gong, Huizhi Yang, Zheng Chai, Zhe Chen, Yuchao Zheng, Qiwei Chen, Feng Zhang, Xun Zhou, Peng Xu, Xiao Yang, Di Wu, Zuotao Liu
|
RankMixer: Scaling Up Ranking Models in Industrial Recommenders
| null | null | null | null |
cs.IR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent progress on large language models (LLMs) has spurred interest in scaling up recommendation systems, yet two practical obstacles remain. First, training and serving cost on industrial Recommenders must respect strict latency bounds and high QPS demands. Second, most human-designed feature-crossing modules in ranking models were inherited from the CPU era and fail to exploit modern GPUs, resulting in low Model Flops Utilization (MFU) and poor scalability. We introduce RankMixer, a hardware-aware model design tailored towards a unified and scalable feature-interaction architecture. RankMixer retains the transformer's high parallelism while replacing quadratic self-attention with multi-head token mixing module for higher efficiency. Besides, RankMixer maintains both the modeling for distinct feature subspaces and cross-feature-space interactions with Per-token FFNs. We further extend it to one billion parameters with a Sparse-MoE variant for higher ROI. A dynamic routing strategy is adapted to address the inadequacy and imbalance of experts training. Experiments show RankMixer's superior scaling abilities on a trillion-scale production dataset. By replacing previously diverse handcrafted low-MFU modules with RankMixer, we boost the model MFU from 4.5\% to 45\%, and scale our ranking model parameters by 100x while maintaining roughly the same inference latency. We verify RankMixer's universality with online A/B tests across two core application scenarios (Recommendation and Advertisement). Finally, we launch 1B Dense-Parameters RankMixer for full traffic serving without increasing the serving cost, which improves user active days by 0.3\% and total in-app usage duration by 1.08\%.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 12:28:55 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 16:19:32 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Zhu",
"Jie",
""
],
[
"Fan",
"Zhifang",
""
],
[
"Zhu",
"Xiaoxie",
""
],
[
"Jiang",
"Yuchen",
""
],
[
"Wang",
"Hangyu",
""
],
[
"Han",
"Xintian",
""
],
[
"Ding",
"Haoran",
""
],
[
"Wang",
"Xinmin",
""
],
[
"Zhao",
"Wenlin",
""
],
[
"Gong",
"Zhen",
""
],
[
"Yang",
"Huizhi",
""
],
[
"Chai",
"Zheng",
""
],
[
"Chen",
"Zhe",
""
],
[
"Zheng",
"Yuchao",
""
],
[
"Chen",
"Qiwei",
""
],
[
"Zhang",
"Feng",
""
],
[
"Zhou",
"Xun",
""
],
[
"Xu",
"Peng",
""
],
[
"Yang",
"Xiao",
""
],
[
"Wu",
"Di",
""
],
[
"Liu",
"Zuotao",
""
]
] |
2507.15611
|
Vo Phuc Dang
|
Dang Vo Phuc
|
On the computation of $\mathrm{Ext}_{\mathcal A}^{k,k+*}(\mathbb{Z}/2,\mathbb{Z}/2)$ for $k \leq 5$
|
21 pages. Any constructive suggestions are most welcome
| null | null | null |
math.AT math.GT math.RA
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This Note presents a computational algorithm for determining a basis of the cohomology of the mod 2 Steenrod algebra, $\mathrm{Ext}_{\mathcal A}^{k, k+*}(\mathbb{Z}/2, \mathbb{Z}/2)$ for $k \leq 5$, based on the well-known generators and the Adem relations given in Chen's work [2]. The purpose of this algorithm is to verify the hand-computed results for $\mathrm{Ext}_{\mathcal A}^{4, 4+*}(\mathbb{Z}/2, \mathbb{Z}/2)$ presented in our corrected papers [8, 9, 10]. Combining our most recent works [11, 12], the verification of the domain and the codomain of the fourth algebraic transfer in specific degrees can now be completed.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 13:34:59 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Jul 2025 23:49:05 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Phuc",
"Dang Vo",
""
]
] |
2507.15744
|
Abdelhakim Necir Necir
|
Saida Mancer, Abdelhakim Necir, Djamel Meraghni
|
Robust and Smooth Estimation of the Extreme Tail Index via Weighted Minimum Density Power Divergence
| null | null | null | null |
math.ST stat.TH
|
http://creativecommons.org/licenses/by/4.0/
|
By introducing a weight function into the density power divergence, we develop a new class of robust and smooth estimators for the tail index of Pareto-type distributions, offering improved efficiency in the presence of outliers. These estimators can be viewed as a robust generalization of both weighted least squares and kernel-based tail index estimators. We establish the consistency and asymptotic normality of the proposed class. A simulation study is conducted to assess their finite-sample performance in comparison with existing methods.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 15:58:52 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 05:49:13 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Mancer",
"Saida",
""
],
[
"Necir",
"Abdelhakim",
""
],
[
"Meraghni",
"Djamel",
""
]
] |
2507.15765
|
Feng-Qi Cui
|
Feng-Qi Cui, Anyang Tong, Jinyang Huang, Jie Zhang, Dan Guo, Zhi Liu, Meng Wang
|
Learning from Heterogeneity: Generalizing Dynamic Facial Expression Recognition via Distributionally Robust Optimization
|
Accepted by ACM MM'25
| null | null | null |
cs.CV cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Dynamic Facial Expression Recognition (DFER) plays a critical role in affective computing and human-computer interaction. Although existing methods achieve comparable performance, they inevitably suffer from performance degradation under sample heterogeneity caused by multi-source data and individual expression variability. To address these challenges, we propose a novel framework, called Heterogeneity-aware Distributional Framework (HDF), and design two plug-and-play modules to enhance time-frequency modeling and mitigate optimization imbalance caused by hard samples. Specifically, the Time-Frequency Distributional Attention Module (DAM) captures both temporal consistency and frequency robustness through a dual-branch attention design, improving tolerance to sequence inconsistency and visual style shifts. Then, based on gradient sensitivity and information bottleneck principles, an adaptive optimization module Distribution-aware Scaling Module (DSM) is introduced to dynamically balance classification and contrastive losses, enabling more stable and discriminative representation learning. Extensive experiments on two widely used datasets, DFEW and FERV39k, demonstrate that HDF significantly improves both recognition accuracy and robustness. Our method achieves superior weighted average recall (WAR) and unweighted average recall (UAR) while maintaining strong generalization across diverse and imbalanced scenarios. Codes are released at https://github.com/QIcita/HDF_DFER.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 16:21:47 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Cui",
"Feng-Qi",
""
],
[
"Tong",
"Anyang",
""
],
[
"Huang",
"Jinyang",
""
],
[
"Zhang",
"Jie",
""
],
[
"Guo",
"Dan",
""
],
[
"Liu",
"Zhi",
""
],
[
"Wang",
"Meng",
""
]
] |
2507.15857
|
Mihir Prabhudesai
|
Mihir Prabhudesai, Menging Wu, Amir Zadeh, Katerina Fragkiadaki, Deepak Pathak
|
Diffusion Beats Autoregressive in Data-Constrained Settings
|
Project Webpage: https://diffusion-scaling.github.io
| null | null | null |
cs.LG cs.AI cs.CV cs.RO
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Autoregressive (AR) models have long dominated the landscape of large language models, driving progress across a wide range of tasks. Recently, diffusion-based language models have emerged as a promising alternative, though their advantages over AR models remain underexplored. In this paper, we systematically study masked diffusion models in data-constrained settings-where training involves repeated passes over limited data-and find that they significantly outperform AR models when compute is abundant but data is scarce. Diffusion models make better use of repeated data, achieving lower validation loss and superior downstream performance. We interpret this advantage as implicit data augmentation: masked diffusion exposes the model to a diverse distribution of token orderings and prediction tasks, unlike AR's fixed left-to-right factorization. We find new scaling laws for diffusion models and derive a closed-form expression for the critical compute threshold at which diffusion begins to outperform AR. These results suggest that when data, not compute, is the bottleneck, diffusion models offer a compelling alternative to the standard AR paradigm. Our code is available at: https://diffusion-scaling.github.io.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 17:59:57 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 17:55:24 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Prabhudesai",
"Mihir",
""
],
[
"Wu",
"Menging",
""
],
[
"Zadeh",
"Amir",
""
],
[
"Fragkiadaki",
"Katerina",
""
],
[
"Pathak",
"Deepak",
""
]
] |
2507.16068
|
Zhehui Huang
|
Zhehui Huang, Guangyao Shi, Yuwei Wu, Vijay Kumar, and Gaurav S. Sukhatme
|
Compositional Coordination for Multi-Robot Teams with Large Language Models
|
9 pages, 4 figures
| null | null | null |
cs.RO cs.AI cs.LG cs.MA
|
http://creativecommons.org/licenses/by/4.0/
|
Multi-robot coordination has traditionally relied on a mission-specific and expert-driven pipeline, where natural language mission descriptions are manually translated by domain experts into mathematical formulation, algorithm design, and executable code. This conventional process is labor-intensive, inaccessible to non-experts, and inflexible to changes in mission requirements. Here, we propose LAN2CB (Language to Collective Behavior), a novel framework that leverages large language models (LLMs) to streamline and generalize the multi-robot coordination pipeline. LAN2CB transforms natural language (NL) mission descriptions into executable Python code for multi-robot systems through two core modules: (1) Mission Analysis, which parses mission descriptions into behavior trees, and (2) Code Generation, which leverages the behavior tree and a structured knowledge base to generate robot control code. We further introduce a dataset of natural language mission descriptions to support development and benchmarking. Experiments in both simulation and real-world environments demonstrate that LAN2CB enables robust and flexible multi-robot coordination from natural language, significantly reducing manual engineering effort and supporting broad generalization across diverse mission types. Website: https://sites.google.com/view/lan-cb
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 21:09:15 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 09:25:12 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Huang",
"Zhehui",
""
],
[
"Shi",
"Guangyao",
""
],
[
"Wu",
"Yuwei",
""
],
[
"Kumar",
"Vijay",
""
],
[
"Sukhatme",
"Gaurav S.",
""
]
] |
2507.16088
|
Alexandra Junell Brown
|
Alexandra Junell, Argyro Sasli, Felipe Fontinele Nunes, Maojie Xu, Benny Border, Nabeel Rehemtulla, Mariia Rizhko, Yu-Jing Qin, Theophile Jegou Du Laz, Antoine Le Calloch, Sushant Sharma Chaudhary, Shaowei Wu, Jesper Sollerman, Niharika Sravan, Steven L. Groom, David Hale, Mansi M. Kasliwal, Josiah Purdum, Avery Wold, Matthew J. Graham, Michael W. Coughlin
|
Applying multimodal learning to Classify transient Detections Early (AppleCiDEr) I: Data set, methods, and infrastructure
|
17 pages, 8 figures
| null | null | null |
astro-ph.IM astro-ph.HE
|
http://creativecommons.org/licenses/by/4.0/
|
Modern time-domain surveys like the Zwicky Transient Facility (ZTF) and the Legacy Survey of Space and Time (LSST) generate hundreds of thousands to millions of alerts, demanding automatic, unified classification of transients and variable stars for efficient follow-up. We present AppleCiDEr (Applying Multimodal Learning to Classify Transient Detections Early), a novel framework that integrates four key data modalities (photometry, image cutouts, metadata, and spectra) to overcome limitations of single-modality classification approaches. Our architecture introduces (i) two transformer encoders for photometry, (ii) a multimodal convolutional neural network (CNN) with domain-specialized metadata towers and Mixture-of-Experts fusion for combining metadata and images, and (iii) a CNN for spectra classification. Training on ~ 30,000 real ZTF alerts, AppleCiDEr achieves high accuracy, allowing early identification and suggesting follow-up for rare transient spectra. The system provides the first unified framework for both transient and variable star classification using real observational data, with seamless integration into brokering pipelines, demonstrating readiness for the LSST era.
|
[
{
"version": "v1",
"created": "Mon, 21 Jul 2025 21:55:14 GMT"
},
{
"version": "v2",
"created": "Thu, 24 Jul 2025 14:46:14 GMT"
}
] | 2025-07-25T00:00:00 |
[
[
"Junell",
"Alexandra",
""
],
[
"Sasli",
"Argyro",
""
],
[
"Nunes",
"Felipe Fontinele",
""
],
[
"Xu",
"Maojie",
""
],
[
"Border",
"Benny",
""
],
[
"Rehemtulla",
"Nabeel",
""
],
[
"Rizhko",
"Mariia",
""
],
[
"Qin",
"Yu-Jing",
""
],
[
"Laz",
"Theophile Jegou Du",
""
],
[
"Calloch",
"Antoine Le",
""
],
[
"Chaudhary",
"Sushant Sharma",
""
],
[
"Wu",
"Shaowei",
""
],
[
"Sollerman",
"Jesper",
""
],
[
"Sravan",
"Niharika",
""
],
[
"Groom",
"Steven L.",
""
],
[
"Hale",
"David",
""
],
[
"Kasliwal",
"Mansi M.",
""
],
[
"Purdum",
"Josiah",
""
],
[
"Wold",
"Avery",
""
],
[
"Graham",
"Matthew J.",
""
],
[
"Coughlin",
"Michael W.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.