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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2508.18799
|
Hassan Abid
|
Hassan Abid, Khan Muhammad, Muhammad Haris Khan
|
Robust and Label-Efficient Deep Waste Detection
|
Accepted at BMVC 2025
| null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Effective waste sorting is critical for sustainable recycling, yet AI research in this domain continues to lag behind commercial systems due to limited datasets and reliance on legacy object detectors. In this work, we advance AI-driven waste detection by establishing strong baselines and introducing an ensemble-based semi-supervised learning framework. We first benchmark state-of-the-art Open-Vocabulary Object Detection (OVOD) models on the real-world ZeroWaste dataset, demonstrating that while class-only prompts perform poorly, LLM-optimized prompts significantly enhance zero-shot accuracy. Next, to address domain-specific limitations, we fine-tune modern transformer-based detectors, achieving a new baseline of 51.6 mAP. We then propose a soft pseudo-labeling strategy that fuses ensemble predictions using spatial and consensus-aware weighting, enabling robust semi-supervised training. Applied to the unlabeled ZeroWaste-s subset, our pseudo-annotations achieve performance gains that surpass fully supervised training, underscoring the effectiveness of scalable annotation pipelines. Our work contributes to the research community by establishing rigorous baselines, introducing a robust ensemble-based pseudo-labeling pipeline, generating high-quality annotations for the unlabeled ZeroWaste-s subset, and systematically evaluating OVOD models under real-world waste sorting conditions. Our code is available at: https://github.com/h-abid97/robust-waste-detection.
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 08:34:04 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 10:07:31 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Abid",
"Hassan",
""
],
[
"Muhammad",
"Khan",
""
],
[
"Khan",
"Muhammad Haris",
""
]
] |
2508.18992
|
Artur Khairullin
|
Ernest A. Dyagin and Nikita I. Kulin and Artur R. Khairullin and Viktor N. Zhuravlev and Alena N. Sitkina
|
Automatic Prompt Optimization with Prompt Distillation
| null | null | null | null |
cs.CL cs.AI cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Autoprompting is the process of automatically selecting optimized prompts for language models, which is gaining popularity due to the rapid development of prompt engineering driven by extensive research in the field of large language models (LLMs). This paper presents DistillPrompt -- a novel autoprompting method based on large language models that employs a multi-stage integration of task-specific information into prompts using training data. DistillPrompt utilizes distillation, compression, and aggregation operations to explore the prompt space more thoroughly. The method was tested on different datasets for text classification and generation tasks using the t-lite-instruct-0.1 language model. The results demonstrate a significant average improvement (e.g., 20.12% across the entire dataset compared to Grips) in key metrics over existing methods in the field, establishing DistillPrompt as one of the most effective non-gradient approaches in autoprompting.
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 12:46:58 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 15:50:16 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Dyagin",
"Ernest A.",
""
],
[
"Kulin",
"Nikita I.",
""
],
[
"Khairullin",
"Artur R.",
""
],
[
"Zhuravlev",
"Viktor N.",
""
],
[
"Sitkina",
"Alena N.",
""
]
] |
2508.19005
|
Jie Zhou
|
Yuxuan Cai, Yipeng Hao, Jie Zhou, Hang Yan, Zhikai Lei, Rui Zhen, Zhenhua Han, Yutao Yang, Junsong Li, Qianjun Pan, Tianyu Huai, Qin Chen, Xin Li, Kai Chen, Bo Zhang, Xipeng Qiu, and Liang He
|
Building Self-Evolving Agents via Experience-Driven Lifelong Learning: A Framework and Benchmark
| null | null | null | null |
cs.AI cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
As AI advances toward general intelligence, the focus is shifting from systems optimized for static tasks to creating open-ended agents that learn continuously. In this paper, we introduce Experience-driven Lifelong Learning (ELL), a framework for building self-evolving agents capable of continuous growth through real-world interaction. The framework is built on four core principles: (1) Experience Exploration: Agents learn through continuous, self-motivated interaction with dynamic environments, navigating interdependent tasks and generating rich experiential trajectories. (2) Long-term Memory: Agents preserve and structure historical knowledge, including personal experiences, domain expertise, and commonsense reasoning, into a persistent memory system. (3) Skill Learning: Agents autonomously improve by abstracting recurring patterns from experience into reusable skills, which are actively refined and validated for application in new tasks. (4) Knowledge Internalization: Agents internalize explicit and discrete experiences into implicit and intuitive capabilities as "second nature".
We also introduce StuLife, a benchmark dataset for ELL that simulates a student's holistic college journey, from enrollment to academic and personal development, across three core phases and ten detailed sub-scenarios. StuLife is designed around three key paradigm
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 13:04:28 GMT"
},
{
"version": "v2",
"created": "Tue, 02 Sep 2025 03:28:14 GMT"
},
{
"version": "v3",
"created": "Sat, 06 Sep 2025 08:21:18 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Cai",
"Yuxuan",
""
],
[
"Hao",
"Yipeng",
""
],
[
"Zhou",
"Jie",
""
],
[
"Yan",
"Hang",
""
],
[
"Lei",
"Zhikai",
""
],
[
"Zhen",
"Rui",
""
],
[
"Han",
"Zhenhua",
""
],
[
"Yang",
"Yutao",
""
],
[
"Li",
"Junsong",
""
],
[
"Pan",
"Qianjun",
""
],
[
"Huai",
"Tianyu",
""
],
[
"Chen",
"Qin",
""
],
[
"Li",
"Xin",
""
],
[
"Chen",
"Kai",
""
],
[
"Zhang",
"Bo",
""
],
[
"Qiu",
"Xipeng",
""
],
[
"He",
"Liang",
""
]
] |
2508.19026
|
Gueter Josmy Faure
|
Gueter Josmy Faure, Min-Hung Chen, Jia-Fong Yeh, Ying Cheng, Hung-Ting Su, Yung-Hao Tang, Shang-Hong Lai, Winston H. Hsu
|
MovieCORE: COgnitive REasoning in Movies
|
Accepted for EMNLP'2025 Main Conference. Project Page: https://joslefaure.github.io/assets/html/moviecore.html
| null | null | null |
cs.CL cs.AI cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
This paper introduces MovieCORE, a novel video question answering (VQA) dataset designed to probe deeper cognitive understanding of movie content. Unlike existing datasets that focus on surface-level comprehension, MovieCORE emphasizes questions that engage System-2 thinking while remaining specific to the video material. We present an innovative agentic brainstorming approach, utilizing multiple large language models (LLMs) as thought agents to generate and refine high-quality question-answer pairs. To evaluate dataset quality, we develop a set of cognitive tests assessing depth, thought-provocation potential, and syntactic complexity. We also propose a comprehensive evaluation scheme for assessing VQA model performance on deeper cognitive tasks. To address the limitations of existing video-language models (VLMs), we introduce an agentic enhancement module, Agentic Choice Enhancement (ACE), which improves model reasoning capabilities post-training by up to 25%. Our work contributes to advancing movie understanding in AI systems and provides valuable insights into the capabilities and limitations of current VQA models when faced with more challenging, nuanced questions about cinematic content. Our project page, dataset and code can be found at https://joslefaure.github.io/assets/html/moviecore.html.
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 13:43:45 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 15:33:50 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Faure",
"Gueter Josmy",
""
],
[
"Chen",
"Min-Hung",
""
],
[
"Yeh",
"Jia-Fong",
""
],
[
"Cheng",
"Ying",
""
],
[
"Su",
"Hung-Ting",
""
],
[
"Tang",
"Yung-Hao",
""
],
[
"Lai",
"Shang-Hong",
""
],
[
"Hsu",
"Winston H.",
""
]
] |
2508.19106
|
Sourav Pal
|
Sourav Pal, Debanjan Sarkar, Alejandro Aviles
|
Tracing Signatures of Modified Gravity in Redshift-Space Galaxy Bispectrum Multipoles: Prospects for Euclid
|
20 pages, 11 figures. Comments are welcome
| null | null | null |
astro-ph.CO gr-qc
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We study the galaxy bispectrum multipoles in the Hu-Sawicki $f(R)$ gravity model, where a scalar degree of freedom mediates a fifth force that is screened in high-density environments. The model is specified by $f_{R0}$, the present-day background value of the scalar field, which controls the strength of deviations from General Relativity (GR). Using perturbation theory, we compute the redshift-space galaxy bispectrum with the full scale- and time-dependent second-order kernels, incorporating corrections from the scale-dependent growth rate and nonlinear screening. Expanding the bispectrum in spherical harmonics, we analyze the sensitivity of the multipoles to modified gravity and forecast their detectability in a \textit{Euclid}-like survey. The monopole ($B_0^0$) and quadrupole ($B_2^0$) show the strongest signatures, with relative deviations of $2\%$-$8\%$ at $z=0.7$ and $k_1\simeq0.3\,h\,{\rm Mpc}^{-1}$ (largest side of the triangle) for $f_{R0}=10^{-5}$. Higher multipoles provide weaker but complementary signals. For \textit{Euclid}, we forecast signal-to-noise ratios up to $\sim30$ for the monopole and $\sim15$ for the quadrupole including the Finger-of-God damping and shot noise effect. These results demonstrate that bispectrum multipoles are a powerful probe of gravity, capable of breaking degeneracies with bias and velocity effects and strengthening constraints on deviations from $\Lambda$CDM.
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 15:10:00 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 05:28:30 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Pal",
"Sourav",
""
],
[
"Sarkar",
"Debanjan",
""
],
[
"Aviles",
"Alejandro",
""
]
] |
2508.19153
|
Xiaowen Tao
|
Yinuo Wang and Gavin Tao
|
QuadKAN: KAN-Enhanced Quadruped Motion Control via End-to-End Reinforcement Learning
|
14pages, 9 figures, Journal paper
| null | null | null |
cs.RO cs.AI cs.CV cs.SY eess.IV eess.SY
|
http://creativecommons.org/publicdomain/zero/1.0/
|
We address vision-guided quadruped motion control with reinforcement learning (RL) and highlight the necessity of combining proprioception with vision for robust control. We propose QuadKAN, a spline-parameterized cross-modal policy instantiated with Kolmogorov-Arnold Networks (KANs). The framework incorporates a spline encoder for proprioception and a spline fusion head for proprioception-vision inputs. This structured function class aligns the state-to-action mapping with the piecewise-smooth nature of gait, improving sample efficiency, reducing action jitter and energy consumption, and providing interpretable posture-action sensitivities. We adopt Multi-Modal Delay Randomization (MMDR) and perform end-to-end training with Proximal Policy Optimization (PPO). Evaluations across diverse terrains, including both even and uneven surfaces and scenarios with static or dynamic obstacles, demonstrate that QuadKAN achieves consistently higher returns, greater distances, and fewer collisions than state-of-the-art (SOTA) baselines. These results show that spline-parameterized policies offer a simple, effective, and interpretable alternative for robust vision-guided locomotion. A repository will be made available upon acceptance.
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 16:05:32 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 01:11:35 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Wang",
"Yinuo",
""
],
[
"Tao",
"Gavin",
""
]
] |
2508.19216
|
Salvador L\'opez-Mart\'inez
|
Salvador L\'opez-Mart\'inez
|
Momentum-mass normalized dark-bright solitons to one dimensional Gross-Pitaevskii systems
| null | null | null | null |
math.AP
|
http://creativecommons.org/licenses/by/4.0/
|
We rigorously establish the existence of dark-bright solitons as traveling wave solutions to a one dimensional defocusing Gross-Pitaevskii system, a widely used model for describing mixtures of Bose-Einstein condensates and nonlinear optical systems. These solitons are shown to exhibit symmetry and radial monotonicity in modulus, and to propagate at subsonic speed. Our method relies on minimizing an energy functional subject to two constraints: the mass of the bright component and a modified momentum of the dark component. The compactness of minimizing sequences is obtained via a concentration-compactness argument, which requires some novel estimates based on symmetric decreasing rearrangements.
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 17:27:31 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 10:18:43 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"López-Martínez",
"Salvador",
""
]
] |
2508.19256
|
Rashid Mushkani
|
Rashid Mushkani, Hugo Berard, Shin Koseki
|
WeDesign: Generative AI-Facilitated Community Consultations for Urban Public Space Design
| null | null | null | null |
cs.HC cs.CY
|
http://creativecommons.org/licenses/by/4.0/
|
Community consultations are integral to urban planning processes intended to incorporate diverse stakeholder perspectives. However, limited resources, visual and spoken language barriers, and uneven power dynamics frequently constrain inclusive decision-making. This paper examines how generative text-to-image methods, specifically Stable Diffusion XL integrated into a custom platform (WeDesign), may support equitable consultations. A half-day workshop in Montreal involved five focus groups, each consisting of architects, urban designers, AI specialists, and residents from varied demographic groups. Additional data was gathered through semi-structured interviews with six urban planning professionals. Participants indicated that immediate visual outputs facilitated creativity and dialogue, yet noted issues in visualizing specific needs of marginalized groups, such as participants with reduced mobility, accurately depicting local architectural elements, and accommodating bilingual prompts. Participants recommended the development of an open-source platform incorporating in-painting tools, multilingual support, image voting functionalities, and preference indicators. The results indicate that generative AI can broaden participation and enable iterative interactions but requires structured facilitation approaches. The findings contribute to discussions on generative AI's role and limitations in participatory urban design.
|
[
{
"version": "v1",
"created": "Wed, 13 Aug 2025 16:55:12 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 01:28:25 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Mushkani",
"Rashid",
""
],
[
"Berard",
"Hugo",
""
],
[
"Koseki",
"Shin",
""
]
] |
2508.19268
|
Xue Han
|
Qing Wang, Xue Han, Jiahui Wang, Lehao Xing, Qian Hu, Lianlian Zhang, Chao Deng, Junlan Feng
|
MultiPL-MoE: Multi-Programming-Lingual Extension of Large Language Models through Hybrid Mixture-of-Experts
| null | null | null | null |
cs.CL cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Despite LLMs' excellent code creation capabilities, multilingual code generation remains extremely challenging. To address this, we intent to improve the multi-programming-lingual (MultiPL) performance of the base LLMs while retaining the most popular ones using restricted computational resources. We consider MultiPL to be a special case of multiple natural languages and propose a MultiPL extension of LLMs utilizing a hybrid mixture of experts (MoE), called MultiPL-MoE. Specifically, MultiPL-MoE combines two paired MoEs to optimize expert selection at both the token and segment levels. The token-level MoE is a standard upcycling MoE structure with a shared expert and a novel gate weight normalization approach that aids in the final fusion with the segment-level MoE. The segment-level MoE incorporates two innovative designs to better capture the syntactic structure and contextual patterns of programming languages: First, using a sliding window to partition the input token sequence into multiple segments; Then, adopting an expert-choice routing strategy that allows experts to select the top-k segments. The results of the experiment proved the effectiveness of MultiPL-MoE.
|
[
{
"version": "v1",
"created": "Fri, 22 Aug 2025 06:24:52 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 08:30:07 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Wang",
"Qing",
""
],
[
"Han",
"Xue",
""
],
[
"Wang",
"Jiahui",
""
],
[
"Xing",
"Lehao",
""
],
[
"Hu",
"Qian",
""
],
[
"Zhang",
"Lianlian",
""
],
[
"Deng",
"Chao",
""
],
[
"Feng",
"Junlan",
""
]
] |
2508.19576
|
Sining Zhoubian
|
Sining Zhoubian, Dan Zhang, Jie Tang
|
ReST-RL: Achieving Accurate Code Reasoning of LLMs with Optimized Self-Training and Decoding
|
21 pages, 4 figures
| null | null | null |
cs.AI cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
With respect to improving the reasoning accuracy of LLMs, the representative reinforcement learning (RL) method GRPO faces failure due to insignificant reward variance, while verification methods based on process reward models (PRMs) suffer from difficulties with training data acquisition and verification effectiveness. To tackle these problems, this paper introduces ReST-RL, a unified LLM RL paradigm that significantly improves LLM's code reasoning ability by combining an improved GRPO algorithm with a meticulously designed test time decoding method assisted by a value model (VM). As the first stage of policy reinforcement, ReST-GRPO adopts an optimized ReST algorithm to filter and assemble high-value training data, increasing the reward variance of GRPO sampling, thus improving the effectiveness and efficiency of training. After the basic reasoning ability of LLM policy has been improved, we further propose a test time decoding optimization method called VM-MCTS. Through Monte-Carlo Tree Search (MCTS), we collect accurate value targets with no annotation required, on which VM training is based. When decoding, the VM is deployed by an adapted MCTS algorithm to provide precise process signals as well as verification scores, assisting the LLM policy to achieve high reasoning accuracy. We conduct extensive experiments on coding problems to verify the validity of the proposed RL paradigm. Upon comparison, our approach significantly outperforms other reinforcement training baselines (e.g., naive GRPO and ReST-DPO), as well as decoding and verification baselines (e.g., PRM-BoN and ORM-MCTS) on well-known coding benchmarks of various levels (e.g., APPS, BigCodeBench, and HumanEval), indicating its power to strengthen the reasoning ability of LLM policies. Codes for our project can be found at https://github.com/THUDM/ReST-RL.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 05:16:03 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 13:12:19 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Zhoubian",
"Sining",
""
],
[
"Zhang",
"Dan",
""
],
[
"Tang",
"Jie",
""
]
] |
2508.19597
|
Zirui Li
|
Zirui Li, Yunlong Lin, Guodong Du, Xiaocong Zhao, Cheng Gong, Chen Lv, Chao Lu, Jianwei Gong
|
Complementary Learning System Empowers Online Continual Learning of Vehicle Motion Forecasting in Smart Cities
|
19 pages, 6 figures
| null | null | null |
cs.LG cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Artificial intelligence underpins most smart city services, yet deep neural network (DNN) that forecasts vehicle motion still struggle with catastrophic forgetting, the loss of earlier knowledge when models are updated. Conventional fixes enlarge the training set or replay past data, but these strategies incur high data collection costs, sample inefficiently and fail to balance long- and short-term experience, leaving them short of human-like continual learning. Here we introduce Dual-LS, a task-free, online continual learning paradigm for DNN-based motion forecasting that is inspired by the complementary learning system of the human brain. Dual-LS pairs two synergistic memory rehearsal replay mechanisms to accelerate experience retrieval while dynamically coordinating long-term and short-term knowledge representations. Tests on naturalistic data spanning three countries, over 772,000 vehicles and cumulative testing mileage of 11,187 km show that Dual-LS mitigates catastrophic forgetting by up to 74.31\% and reduces computational resource demand by up to 94.02\%, markedly boosting predictive stability in vehicle motion forecasting without inflating data requirements. Meanwhile, it endows DNN-based vehicle motion forecasting with computation efficient and human-like continual learning adaptability fit for smart cities.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 06:19:21 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 09:35:56 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Li",
"Zirui",
""
],
[
"Lin",
"Yunlong",
""
],
[
"Du",
"Guodong",
""
],
[
"Zhao",
"Xiaocong",
""
],
[
"Gong",
"Cheng",
""
],
[
"Lv",
"Chen",
""
],
[
"Lu",
"Chao",
""
],
[
"Gong",
"Jianwei",
""
]
] |
2508.19710
|
Joshua Tin Yau Tse
|
Joshua T. Y. Tse, Taisuke Enomoto, Shunsuke Murai, Katsuhisa Tanaka
|
Modelling Purcell enhancement of metasurfaces supporting quasi-bound states in the continuum
|
26 pages, 10 figures
| null | null | null |
physics.optics
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Bound states in the continuum (BIC) exhibit extremely high quality factors due to the lack of radiation loss and thus are widely studied for Purcell enhancement. However, a closer examination reveals that the enhancement is absent at the BIC due to the lack of out-coupling capability, but the strong enhancement is only observed at nearby configuration, namely quasi-BIC. To study this unique behavior of the Purcell enhancement near BIC, we built an analytically model with spectral parameters to analyze the Purcell enhancement on metasurfaces supporting quasi-BIC. Our analytical model predicts the average Purcell enhancement by metasurfaces coupled to a luminescent medium, utilizing parameters that are formulated through the temporal coupled-mode theory and can be derived from measured spectra such as transmissivity and reflectivity. We analyzed several metasurfaces supporting quasi-BIC numerically and experimentally to study the behavior of the spectral parameters as well as the resultant Purcell enhancement. We formulated the interdependence between the quality factor and the out-coupling efficiency, and revealed the existence of optimal detuning from the BIC. We also discovered that our findings are general and applicable towards realistic metasurfaces that are lossy and/or asymmetric. This discovery provides an intuitive model to understand the modal qualities of quasi-BIC and will facilitate optimization of quasi-BIC for luminescence enhancement applications.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 09:17:44 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 02:33:50 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Tse",
"Joshua T. Y.",
""
],
[
"Enomoto",
"Taisuke",
""
],
[
"Murai",
"Shunsuke",
""
],
[
"Tanaka",
"Katsuhisa",
""
]
] |
2508.19769
|
Shu Shen
|
Shu Shen, C. L. Philip Chen, Tong Zhang
|
AIM: Adaptive Intra-Network Modulation for Balanced Multimodal Learning
|
13pages,7 figures
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Multimodal learning has significantly enhanced machine learning performance but still faces numerous challenges and limitations. Imbalanced multimodal learning is one of the problems extensively studied in recent works and is typically mitigated by modulating the learning of each modality. However, we find that these methods typically hinder the dominant modality's learning to promote weaker modalities, which affects overall multimodal performance. We analyze the cause of this issue and highlight a commonly overlooked problem: optimization bias within networks. To address this, we propose Adaptive Intra-Network Modulation (AIM) to improve balanced modality learning. AIM accounts for differences in optimization state across parameters and depths within the network during modulation, achieving balanced multimodal learning without hindering either dominant or weak modalities for the first time. Specifically, AIM decouples the dominant modality's under-optimized parameters into Auxiliary Blocks and encourages reliance on these performance-degraded blocks for joint training with weaker modalities. This approach effectively prevents suppression of weaker modalities while enabling targeted optimization of under-optimized parameters to improve the dominant modality. Additionally, AIM assesses modality imbalance level across network depths and adaptively adjusts modulation strength at each depth. Experimental results demonstrate that AIM outperforms state-of-the-art imbalanced modality learning methods across multiple benchmarks and exhibits strong generalizability across different backbones, fusion strategies, and optimizers.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 10:53:36 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 03:08:42 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Shen",
"Shu",
""
],
[
"Chen",
"C. L. Philip",
""
],
[
"Zhang",
"Tong",
""
]
] |
2508.20144
|
Julio Zanon Diaz
|
Julio Zanon Diaz, Tommy Brennan, Peter Corcoran
|
Navigating the EU AI Act: Foreseeable Challenges in Qualifying Deep Learning-Based Automated Inspections of Class III Medical Devices
|
Critical Review article
| null | null | null |
cs.CY cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
As deep learning (DL) technologies advance, their application in automated visual inspection for Class III medical devices offers significant potential to enhance quality assurance and reduce human error. However, the adoption of such AI-based systems introduces new regulatory complexities-particularly under the EU Artificial Intelligence (AI) Act, which imposes high-risk system obligations that differ in scope and depth from established regulatory frameworks such as the Medical Device Regulation (MDR) and the U.S. FDA Quality System Regulation (QSR). This paper presents a high-level technical assessment of the foreseeable challenges that manufacturers are likely to encounter when qualifying DL-based automated inspections -- specifically static models -- within the existing medical device compliance landscape. It examines divergences in risk management principles, dataset governance, model validation, explainability requirements, and post-deployment monitoring obligations. The discussion also explores potential implementation strategies and highlights areas of uncertainty, including data retention burdens, global compliance implications, and the practical difficulties of achieving statistical significance in validation with limited defect data. Disclaimer: This paper presents a technical perspective and does not constitute legal or regulatory advice.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 09:17:03 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 15:48:19 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Diaz",
"Julio Zanon",
""
],
[
"Brennan",
"Tommy",
""
],
[
"Corcoran",
"Peter",
""
]
] |
2508.20164
|
Subir Sachdev
|
Pietro M. Bonetti and Maine Christos and Alexander Nikolaenko and Aavishkar A. Patel and Subir Sachdev
|
Critical quantum liquids and the cuprate high temperature superconductors
|
91 pages, 43 figures. Review article based on lectures by SS at Boulder, Trieste, Hong Kong, with links to lecture videos. Comments welcome
| null | null | null |
cond-mat.str-el hep-th
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present a theoretical framework for the cuprate superconductors, rooted in a fractionalized Fermi liquid (FL*) description of the intermediate-temperature pseudogap phase at low doping. The FL* theory predicted hole pockets each of fractional area $p/8$ at hole doping $p$, in contrast to the area $p/4$ in a spin density wave state or its thermal fluctuation. A recent magnetotransport observation of the Yamaji angle is in good agreement with area $p/8$.
We review a theory for the FL* phase of a single-band model using a layer construction with a pair of ancilla qubits on each site: the Ancilla Layer Model (ALM). Its mean field yields hole pockets of area $p/8$, and matches the gapped photoemission spectrum in the anti-nodal region of the Brillouin zone. Fluctuations are described by the SU(2) gauge theory of a background spin liquid with critical Dirac spinons. A Monte Carlo study of the thermal SU(2) gauge theory transforms the hole pockets into Fermi arcs in photoemission. One route to confinement of FL* upon lowering temperature yields a $d$-wave superconductor via a Kosterlitz-Thouless transition of $h/(2e)$ vortices, with nodal Bogoliubov quasiparticles featuring anisotropic velocities and vortices surrounded by charge order halos. An alternative route produces a charge-ordered metallic state that exhibits quantum oscillations in agreement with experiments.
Increasing doping from the FL* phase in the ALM drives a transition to a Fermi liquid at large doping, passing through an intermediate strange metal regime. We formulate a theory of this metal using a critical quantum `charge' liquid of mobile electrons in the presence of disorder, via an extension of the Sachdev-Ye-Kitaev model.
At low temperatures, and across optimal and over doping, we address the regimes of extended non-Fermi liquid behavior by Griffiths effects near quantum phase transitions in disordered metals.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 18:00:01 GMT"
},
{
"version": "v2",
"created": "Sun, 31 Aug 2025 09:35:10 GMT"
},
{
"version": "v3",
"created": "Mon, 08 Sep 2025 16:18:37 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Bonetti",
"Pietro M.",
""
],
[
"Christos",
"Maine",
""
],
[
"Nikolaenko",
"Alexander",
""
],
[
"Patel",
"Aavishkar A.",
""
],
[
"Sachdev",
"Subir",
""
]
] |
2508.20170
|
Ilkham Galiullin
|
Ilkham Galiullin, Antonio C. Rodriguez, Kareem El-Badry, Ilaria Caiazzo, Paula Szkody, Pranav Nagarajan, Samuel Whitebook
|
Optical Spectroscopy of the Most Compact Accreting Binary Harboring a Magnetic White Dwarf and a Hydrogen-rich Donor
|
11 pages, 6 figures, 1 table. Accepted for publication in ApJ Letters
| null |
10.3847/2041-8213/adff82
| null |
astro-ph.SR astro-ph.HE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Accreting white dwarfs in close binary systems, commonly known as cataclysmic variables (CVs), with orbital periods below the canonical period minimum ($\approx$ 80 minutes) are rare. Such short periods can only be reached if the donor star in the CV is either significantly evolved before initiating mass transfer to the white dwarf (WD) or metal-poor. We present optical photometry and spectroscopy of Gaia19bxc, a high-amplitude variable identified as a polar CV with an exceptionally short orbital period of 64.42 minutes - well below the canonical CV period minimum. High-speed photometry confirms persistent double-peaked variability consistent with cyclotron beaming, thus indicating the presence of a magnetic WD. Phase-resolved Keck/LRIS spectroscopy reveals strong hydrogen and helium emission lines but no donor features, indicating the accretor is a magnetic WD and the donor is hydrogen-rich, but cold and faint. The absence of a detectable donor and the low inferred temperature ($\lesssim$ 3500 K) disfavor an evolved donor scenario. Instead, the short period and the system's halo-like kinematics suggest Gaia19bxc may be the first known metal-poor polar. Because metal-poor donors are more compact than solar-metallicity donors of the same mass, they can reach shorter minimum periods. Gaia19bxc is one of only a handful of known metal-poor CVs below the canonical period minimum and has the shortest period of any such magnetic system discovered to date.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 18:00:05 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Galiullin",
"Ilkham",
""
],
[
"Rodriguez",
"Antonio C.",
""
],
[
"El-Badry",
"Kareem",
""
],
[
"Caiazzo",
"Ilaria",
""
],
[
"Szkody",
"Paula",
""
],
[
"Nagarajan",
"Pranav",
""
],
[
"Whitebook",
"Samuel",
""
]
] |
2508.20184
|
Rafael I. Nepomechie
|
Nicolas Cramp\'e, Rafael I. Nepomechie, Luc Vinet, Nabi Zare Harofteh
|
$su(2)$ symmetry of XX spin chains
|
7 pages; v2: references added
| null | null | null |
cond-mat.stat-mech math-ph math.MP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We show that, after suitably adjusting a uniform transverse magnetic field, the generic inhomogeneous open XX spin chain has a two-fold degeneracy, and an exact $su(2)$ symmetry whose "inhomogeneous" nonlocal generators depend on coefficients that can be explicitly computed for models associated with discrete orthogonal polynomials.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 18:04:20 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 00:19:22 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Crampé",
"Nicolas",
""
],
[
"Nepomechie",
"Rafael I.",
""
],
[
"Vinet",
"Luc",
""
],
[
"Harofteh",
"Nabi Zare",
""
]
] |
2508.20212
|
Minghao Hu
|
Minghao Hu, Junzhe Wang, Weisen Zhao, Qiang Zeng, Lannan Luo
|
FlowMalTrans: Unsupervised Binary Code Translation for Malware Detection Using Flow-Adapter Architecture
|
This paper is accepted to EMNLP 2025 Findings
| null | null | null |
cs.CR cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Applying deep learning to malware detection has drawn great attention due to its notable performance. With the increasing prevalence of cyberattacks targeting IoT devices, there is a parallel rise in the development of malware across various Instruction Set Architectures (ISAs). It is thus important to extend malware detection capacity to multiple ISAs. However, training a deep learning-based malware detection model usually requires a large number of labeled malware samples. The process of collecting and labeling sufficient malware samples to build datasets for each ISA is labor-intensive and time-consuming. To reduce the burden of data collection, we propose to leverage the ideas of Neural Machine Translation (NMT) and Normalizing Flows (NFs) for malware detection. Specifically, when dealing with malware in a certain ISA, we translate it to an ISA with sufficient malware samples (like X86-64). This allows us to apply a model trained on one ISA to analyze malware from another ISA. Our approach reduces the data collection effort by enabling malware detection across multiple ISAs using a model trained on a single ISA.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 18:39:54 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 21:31:54 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Hu",
"Minghao",
""
],
[
"Wang",
"Junzhe",
""
],
[
"Zhao",
"Weisen",
""
],
[
"Zeng",
"Qiang",
""
],
[
"Luo",
"Lannan",
""
]
] |
2508.20284
|
Sercan H\"usn\"ugil
|
Sercan H\"usn\"ugil and Luis Lehner
|
Sourced Carrollian Fluids Dual to Black Hole Horizons
|
24 pages, 7 figures; added references for the introduction, increased line widths for the figures 1, 4 & 5
| null | null | null |
gr-qc hep-th
|
http://creativecommons.org/licenses/by/4.0/
|
The (degenerate) geometry of event horizons is linked to Carrollian fluids. We investigate the behavior of event horizons via a perturbative coupling to a massless scalar field, making connections to Carrollian hydrodynamics with a driving source, and discuss the fluid equilibration in tandem with the horizon's relaxation to equilibrium. We observe that after the perturbation dies off, the Carrollian fluid energy and momentum densities approach equilibrium as the horizon asymptotically becomes non-expanding. We connect the equilibration of the Carrollian fluid dual to the black hole horizon through the expansion of its background geometry.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 21:30:17 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 08:33:57 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Hüsnügil",
"Sercan",
""
],
[
"Lehner",
"Luis",
""
]
] |
2508.20299
|
Peter Pronko PhD
|
Peter P. Pronko and Paul A. Van Rompay
|
Magnetic Centrifuge Effects in Ultrafast Laser Ablation Plasmas
|
Replacement version has title change and improved section numbering along with minor clarifications in textual content. The title change was made in favor of a more generalized and simplified version. The abstract and conclusion have remained the same. No changes in data or analysis were made
| null | null | null |
physics.plasm-ph physics.app-ph
|
http://creativecommons.org/publicdomain/zero/1.0/
|
A self-consistent model is developed to explain the anomalously large enrichment of nickel isotopes observed in ablation plumes from ultrafast laser irradiation of solid surfaces. The model is based on the spontaneous creation of a magnetic centrifuge in the ablation plume and the associated cyclotron rotation of plasma ions with rotation rates on the order of 10^9 radians per second. Mass separation occurs around the radial coordinate of cylindrical symmetry with longitudinal axis normal to the ablating surface. A Gaussian shaped radial magnetic field distribution is extracted for Ni isotopes with peak field of 53 Megagauss and average field of 20 Megagauss. In addition to cyclotron rotation of ions, a rigid rotor model is also presented that is associated with the hydrodynamic rotation of the entire plasma and is shown to be of little consequence for the isotope enrichment. Cyclotron rotations dominate the process. A profound resonance of enrichment is observed for specific cyclotron charge states and is shown to be associated with Ion Bernstein Wave (IBW) coupling that provides electrostatic acceleration to these resonant orbits.
|
[
{
"version": "v1",
"created": "Wed, 27 Aug 2025 22:10:22 GMT"
},
{
"version": "v2",
"created": "Sat, 30 Aug 2025 17:42:18 GMT"
},
{
"version": "v3",
"created": "Mon, 08 Sep 2025 17:07:25 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Pronko",
"Peter P.",
""
],
[
"Van Rompay",
"Paul A.",
""
]
] |
2508.20401
|
Alexandre Andre
|
Alexandre Andre, Gauthier Roy, Eva Dyer, Kai Wang
|
Revealing Potential Biases in LLM-Based Recommender Systems in the Cold Start Setting
|
In Proceedings of 2nd Workshop on Evaluating and Applying Recommendation Systems with Large Language Models (EARL) at RecSys 2025 (EARL 2025)
| null | null | null |
cs.IR cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Large Language Models (LLMs) are increasingly used for recommendation tasks due to their general-purpose capabilities. While LLMs perform well in rich-context settings, their behavior in cold-start scenarios, where only limited signals such as age, gender, or language are available, raises fairness concerns because they may rely on societal biases encoded during pretraining. We introduce a benchmark specifically designed to evaluate fairness in zero-context recommendation. Our modular pipeline supports configurable recommendation domains and sensitive attributes, enabling systematic and flexible audits of any open-source LLM. Through evaluations of state-of-the-art models (Gemma 3 and Llama 3.2), we uncover consistent biases across recommendation domains (music, movies, and colleges) including gendered and cultural stereotypes. We also reveal a non-linear relationship between model size and fairness, highlighting the need for nuanced analysis.
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 03:57:13 GMT"
},
{
"version": "v2",
"created": "Fri, 05 Sep 2025 18:12:25 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Andre",
"Alexandre",
""
],
[
"Roy",
"Gauthier",
""
],
[
"Dyer",
"Eva",
""
],
[
"Wang",
"Kai",
""
]
] |
2508.20417
|
Ngoc Mai Thieu
|
Chi Minh Bui, Ngoc Mai Thieu, Van Vinh Nguyen, Jason J.Jung, Khac-Hoai Nam Bui
|
KG-CQR: Leveraging Structured Relation Representations in Knowledge Graphs for Contextual Query Retrieval
|
Accepted at Main EMNLP 2025
| null | null | null |
cs.CL cs.DB
|
http://creativecommons.org/licenses/by/4.0/
|
The integration of knowledge graphs (KGs) with large language models (LLMs) offers significant potential to improve the retrieval phase of retrieval-augmented generation (RAG) systems. In this study, we propose KG-CQR, a novel framework for Contextual Query Retrieval (CQR) that enhances the retrieval phase by enriching the contextual representation of complex input queries using a corpus-centric KG. Unlike existing methods that primarily address corpus-level context loss, KG-CQR focuses on query enrichment through structured relation representations, extracting and completing relevant KG subgraphs to generate semantically rich query contexts. Comprising subgraph extraction, completion, and contextual generation modules, KG-CQR operates as a model-agnostic pipeline, ensuring scalability across LLMs of varying sizes without additional training. Experimental results on RAGBench and MultiHop-RAG datasets demonstrate KG-CQR's superior performance, achieving a 4-6% improvement in mAP and a 2-3% improvement in Recall@25 over strong baseline models. Furthermore, evaluations on challenging RAG tasks such as multi-hop question answering show that, by incorporating KG-CQR, the performance consistently outperforms the existing baseline in terms of retrieval effectiveness
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 04:37:15 GMT"
},
{
"version": "v2",
"created": "Fri, 29 Aug 2025 11:37:41 GMT"
},
{
"version": "v3",
"created": "Sat, 06 Sep 2025 04:31:35 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Bui",
"Chi Minh",
""
],
[
"Thieu",
"Ngoc Mai",
""
],
[
"Nguyen",
"Van Vinh",
""
],
[
"Jung",
"Jason J.",
""
],
[
"Bui",
"Khac-Hoai Nam",
""
]
] |
2508.20526
|
Matthieu Gendrin
|
Matthieu Gendrin, St\'ephane Pateux, Xiaoran Jiang, Th\'eo Ladune, Luce Morin
|
Adam SLAM - the last mile of camera calibration with 3DGS
| null |
ORASIS 2025, ISEN Yncr\'ea Ouest, Jun 2025, Le Croisic, France. hal-05131219
| null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
The quality of the camera calibration is of major importance for evaluating progresses in novel view synthesis, as a 1-pixel error on the calibration has a significant impact on the reconstruction quality. While there is no ground truth for real scenes, the quality of the calibration is assessed by the quality of the novel view synthesis. This paper proposes to use a 3DGS model to fine tune calibration by backpropagation of novel view color loss with respect to the cameras parameters. The new calibration alone brings an average improvement of 0.4 dB PSNR on the dataset used as reference by 3DGS. The fine tuning may be long and its suitability depends on the criticity of training time, but for calibration of reference scenes, such as Mip-NeRF 360, the stake of novel view quality is the most important.
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 08:08:54 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Gendrin",
"Matthieu",
""
],
[
"Pateux",
"Stéphane",
""
],
[
"Jiang",
"Xiaoran",
""
],
[
"Ladune",
"Théo",
""
],
[
"Morin",
"Luce",
""
]
] |
2508.20733
|
Hideki Matsumura
|
Munenori Inagaki, Hideki Matsumura, Masanori Sawa, Yukihiro Uchida
|
On a group of normalized solutions of the higher-dimensional Prouhet--Tarry--Escott problem
|
5 pages
| null | null | null |
math.NT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We elucidate, for the first time, a novel group-theoretic structure that arises from certain solutions of the $n$-dimensional Prouhet--Tarry--Escott problem of degree $2$ and size $n$. We prove that the group is isomorphic to the orthogonal group for a certain quadratic form.
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 12:55:59 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 04:24:45 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Inagaki",
"Munenori",
""
],
[
"Matsumura",
"Hideki",
""
],
[
"Sawa",
"Masanori",
""
],
[
"Uchida",
"Yukihiro",
""
]
] |
2508.21045
|
Zhenhua Liu
|
Zhenhua Liu
|
The Hasse Principle for Geometric Variational Problems: An Illustration via Area-minimizing Submanifolds
|
Supersedes arXiv:2310.19860 and arXiv:2401.18074. Added a section on further analogies with number theory
| null | null | null |
math.DG math.AP math.NT
|
http://creativecommons.org/licenses/by/4.0/
|
The Hasse principle in number theory states that information about integral solutions to Diophantine equations can be pieced together from real solutions and solutions modulo prime powers. We show that the Hasse principle holds for area-minimizing submanifolds: information about area-minimizing submanifolds in integral homology can be fully recovered from those in real homology and mod $n$ homology for all $n\in \mathbb{Z}_{\ge 2}.$ As a consequence we derive several surprising conclusions, including: area-minimizing submanifolds in mod $n$ homology are asymptotically much smoother than expected and area-minimizing submanifolds are not generically calibrated. We conjecture that the Hasse principle holds for all geometric variational problems that can be formulated on chain space over different coeffiicients, e.g., Almgren-Pitts min-max, mean curvature flow, Song's spherical Plateau problem, minimizers of elliptic and other general functionals, etc.
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 17:50:21 GMT"
},
{
"version": "v2",
"created": "Tue, 02 Sep 2025 15:33:45 GMT"
},
{
"version": "v3",
"created": "Mon, 08 Sep 2025 16:03:38 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Liu",
"Zhenhua",
""
]
] |
2508.21086
|
Logan McCarty
|
Logan S. McCarty
|
Quantum-inspired probability metrics define a complete, universal space for statistical learning
|
8 pages, 1 figure, 36-page appendix. v2: Corrected typos, matches journal submission
| null | null | null |
stat.ML cs.LG quant-ph
|
http://creativecommons.org/licenses/by/4.0/
|
Comparing probability distributions is a core challenge across the natural, social, and computational sciences. Existing methods, such as Maximum Mean Discrepancy (MMD), struggle in high-dimensional and non-compact domains. Here we introduce quantum probability metrics (QPMs), derived by embedding probability measures in the space of quantum states: positive, unit-trace operators on a Hilbert space. This construction extends kernel-based methods and overcomes the incompleteness of MMD on non-compact spaces. Viewed as an integral probability metric (IPM), QPMs have dual functions that uniformly approximate all bounded, uniformly continuous functions on $\mathbb{R}^n$, offering enhanced sensitivity to subtle distributional differences in high dimensions. For empirical distributions, QPMs are readily calculated using eigenvalue methods, with analytic gradients suited for learning and optimization. Although computationally more intensive for large sample sizes ($O(n^3)$ vs. $O(n^2)$), QPMs can significantly improve performance as a drop-in replacement for MMD, as demonstrated in a classic generative modeling task. By combining the rich mathematical framework of quantum mechanics with classical probability theory, this approach lays the foundation for powerful tools to analyze and manipulate probability measures.
|
[
{
"version": "v1",
"created": "Tue, 26 Aug 2025 22:41:48 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 15:48:34 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"McCarty",
"Logan S.",
""
]
] |
2508.21110
|
Antonio Pasqua Dr.
|
Antonio Pasqua
|
Power-Law and Logarithmic Entropy-Corrected Ricci Dark Energy in a Non-Flat FRW Universe with Viscous Interaction
| null | null | null | null |
gr-qc
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In this work, we consider the power-law corrected and the logarithmic-corrected versions of the Holographic Dark Energy (HDE) model in a non-flat FRW Universe filled with a viscous Dark Energy (DE) interacting with Dark Matter (DM). We propose to replace the infrared cut-off with the inverse of the Ricci scalar curvature $R$. We obtain the equation of state (EoS) parameter $\omega_{\Lambda}$, the deceleration parameter $q$ and the evolution of energy density parameter $\Omega_{\Lambda} '$ in the presence of interaction between DE and DM for both corrections. We study the correspondence of the power-law entropy corrected Ricci dark energy (PLECRDE) and the logarithmic entropy corrected Ricci dark energy (LECRDE) models with the Generalized Chaplygin Gas (GCG), the Modified Variable Chaplygin Gas (MVCG), the New Modified Chaplygin Gas (NMCG), the Yang-Mills (YM) and the Non Linear Electro-Dynamics (NLED) scalar field models.
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 16:51:24 GMT"
},
{
"version": "v2",
"created": "Mon, 01 Sep 2025 10:42:35 GMT"
},
{
"version": "v3",
"created": "Sun, 07 Sep 2025 10:27:10 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Pasqua",
"Antonio",
""
]
] |
2508.21127
|
Zhaoyu Han
|
Zhaoyu Han, Jonah Herzog-Arbeitman, Qiang Gao, Eslam Khalaf
|
Exact models of chiral flat-band superconductors
| null | null | null | null |
cond-mat.str-el cond-mat.supr-con
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Recent experiments have reported the surprising observation of superconductivity in flavor polarized, nearly flat bands (FBs) of rhombohedral graphene. Motivated by these findings, we introduce a class of models for single-flavor FBs with inversion symmetry, where we show a local attractive interaction between orbitals with opposite parities leads to an exact superconducting ground state. We argue that this model can be relevant to realistic multi-flavor systems including short-range repulsion, since the main effect of such repulsion is to induce flavor polarization leaving possibly attractive residual interaction between different flavorless orbitals. The nature of the pairing is determined by the interplay between the FB quantum geometry and the interaction, and is often topological when the parent FB is so. Interestingly, each such model has two nearly degenerate pairing modes, whose energetic competition can be tuned by a change in the charge transfer gap between the two orbitals or electron density. These modes have the same angular momentum but different pairing amplitude structure and possibly different topology. We show that the superfluid stiffness is proportional to the attractive interaction scale using a combination of analytical variational upper bounds and numerical bootstrap lower bounds. We find empirically that the maximum superfluid stiffness is achieved when the hot spots of quantum geometry in the Brillouin zone are marginally filled.
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 18:01:11 GMT"
},
{
"version": "v2",
"created": "Fri, 05 Sep 2025 19:52:16 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Han",
"Zhaoyu",
""
],
[
"Herzog-Arbeitman",
"Jonah",
""
],
[
"Gao",
"Qiang",
""
],
[
"Khalaf",
"Eslam",
""
]
] |
2508.21262
|
Xikai Shan
|
Xikai Shan, Huan Yang, Shude Mao and Otto A. Hannuksela
|
The degeneracy between microlensing wave effect and precession in strongly lensed gravitational wave
|
Comments are welcome!
| null | null | null |
astro-ph.CO astro-ph.HE gr-qc
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Microlensing induced by the stellar field within a strong lensing galaxy can introduce fluctuations in the waveforms of strongly-lensed gravitational waves. When fitting these signals with templates that do not account for microlensing,possible degeneracies can lead to false evidence of certain intrinsic parameters,resulting in a misinterpretation of the properties of the underlying system. For example,the wave effect of microlensing may mimic spin precessions,as both effects generically induce periodic waveform modulations. Although previous studies suggest that lensing-induced modulations can be distinguished from precession using parameter estimation under a geometric-optic approximation,it does not directly apply for lensed image through stellar fields due to large number of stars involved an wave-optic effects. This study aims to evaluate the degree of degeneracy between the stellar-field microlensing and spin precession,investigating whether microlensing leads false evidence of precession. In other words,to what extent observed SLGWs are identified with false evidence of precession. The main findings are as follows. First,assuming O5 sensitivity and parallel spins for the underlying binary black holes,microlensing-induced false evidence of precession is generally weak(15% of the events show significant evidence,30% if the signal-to-noise ratio doubles). Second,for highly magnified events,about 72% of the population show significant evidence of precession,which could serve as an identification criterion for microlensing. This implies the possibility of strong lensing for the recent event GW231123,which is in the mass gap and also shows evidence of precession. Finally,a moderate to strong positive correlation exists between microlensing strength and precession evidence,more pronounced in Type II SLGWs.
|
[
{
"version": "v1",
"created": "Thu, 28 Aug 2025 23:29:03 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 02:42:20 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Shan",
"Xikai",
""
],
[
"Yang",
"Huan",
""
],
[
"Mao",
"Shude",
""
],
[
"Hannuksela",
"Otto A.",
""
]
] |
2508.21423
|
Kunal Dutta
|
Kunal Dutta
|
Constructive l2-Discrepancy Minimization with Additive Deviations
|
A preliminary version was submitted to a conference
| null | null | null |
cs.DM cs.DS math.PR
|
http://creativecommons.org/licenses/by/4.0/
|
The \emph{signed series} problem in the $\ell_2$ norm asks, given set of vectors $v_1,\ldots,v_n\in \mathbf{R}^d$ having at most unit $\ell_2$ norm, does there always exist a series $(\varepsilon_i)_{i\in [n]}$ of $\pm 1$ signs such that for all $i\in [n]$, $\max_{i\in [n]} \|\sum_{j=1}^i \varepsilon_i v_i\|_2 = O(\sqrt{d})$. A result of Banaszczyk [2012, \emph{Rand. Struct. Alg.}] states that there exist signs $\varepsilon_i\in \{-1,1\},\; i\in [n]$ such that $\max_{i\in [n]} \|\sum_{j=1}^i \varepsilon_i v_i\|_2 = O(\sqrt{d+\log n})$. The best constructive bound known so far is of $O(\sqrt{d\log n})$, by Bansal and Garg [2017, \emph{STOC.}, 2019, \emph{SIAM J. Comput.}]. We give a polynomial-time randomized algorithm to find signs $x(i) \in \{-1,1\},\; i\in [n]$ such that \[ \max_{i\in [n]} \|\sum_{j=1}^i x(i)v_i\|_2 = O(\sqrt{d + \log^2 n}) = O(\sqrt{d}+\log n).\] By the constructive reduction of Harvey and Samadi [\emph{COLT}, 2014], this also yields a constructive bound of $O(\sqrt{d}+\log n)$ for the Steinitz problem in the $\ell_2$-norm. Thus, we algorithmically achieve Banaszczyk's bounds for both problems when $d \geq \log^2n$, which also matches the conjectured bounds. Our algorithm is based on the framework on Bansal and Garg, together with a new analysis involving $(i)$ additional linear and spectral orthogonality constraints during the construction of the covariance matrix of the random walk steps, which allow us to control the quadratic variation in the linear as well as the quadratic components of the discrepancy increment vector, alongwith $(ii)$ a ``Freedman-like" version of the Hanson-Wright concentration inequality, for filtration-dependent sums of subgaussian chaoses.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 08:49:41 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 05:59:52 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Dutta",
"Kunal",
""
]
] |
2508.21474
|
Zhongyu Wang
|
Zhong-Yu Wang and Zheng-Wen Long
|
Prediction of $QQqq\bar{s}$ molecular pentaquarks within the extended local hidden gauge approach
|
11 pages, 6 figures
| null | null | null |
hep-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We investigate hadronic molecular states with the quark contents $ccqq\bar{s}$, $bbqq\bar{s}$, and $bcqq\bar{s}$ $(q=u,d)$ by employing the extended local hidden gauge approach. Considering that the $S$-wave meson-baryon interactions are dominated by vector meson exchange, the coupled channels scattering amplitudes are obtained by solving the Bethe-Salpeter equation in its on-shell form. We find that the poles appearing on the complex Riemann sheet are potential candidates for dynamically generated molecular pentaquark states. The results suggest the existence of a total of fourteen molecular states with quantum numbers $I(J^{P})=0(1/2^{-})$, $0(3/2^{-})$, and $0(5/2^{-})$, which arise from the interactions of the $K^{(*)}\Xi_{cc}^{(*)}$, $K^{(*)}\Xi_{bb}^{(*)}$, $K^{(*)}\Xi_{bc}^{(*)}$, and $K^{(*)}\Xi_{bc}^{'}$ channels, respectively. Their binding energies are calculated to be about $0.1-33$ MeV, and this range depends on the free parameter of the theory. Our research contributes to the spectroscopic studies of hadronic molecular pentaquark states.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 09:58:21 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 10:50:00 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Wang",
"Zhong-Yu",
""
],
[
"Long",
"Zheng-Wen",
""
]
] |
2508.21495
|
Filip Szatkowski
|
Piotr Kubaty, Filip Szatkowski, Metod Jazbec, Bartosz W\'ojcik
|
Failure Prediction Is a Better Performance Proxy for Early-Exit Networks Than Calibration
| null | null | null | null |
cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Early-exit models accelerate inference by attaching internal classifiers to intermediate layers of the network, allowing computation to halt once a prediction meets a predefined exit criterion. Most early-exit methods rely on confidence-based exit strategies, which has motivated prior work to calibrate intermediate classifiers in pursuit of improved performance-efficiency trade-offs. In this paper, we argue that calibration metrics can be misleading indicators of multi-exit model performance. Specifically, we present empirical evidence showing that miscalibrated networks can outperform calibrated ones. As an alternative, we propose using failure prediction as a more informative proxy for early-exit model performance. Unlike calibration, failure prediction captures changes in sample rankings and correlates strongly with efficiency gains, offering a more reliable framework for designing and evaluating early-exit models.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 10:21:46 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 18:29:33 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Kubaty",
"Piotr",
""
],
[
"Szatkowski",
"Filip",
""
],
[
"Jazbec",
"Metod",
""
],
[
"Wójcik",
"Bartosz",
""
]
] |
2508.21590
|
Alexander Popov
|
Alexander D. Popov
|
Dirac particles, spin and photons
|
56 pages; v2: minor corrections
| null | null | null |
hep-th math-ph math.MP quant-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We describe relativistic particles with spin as points moving in phase space $X=T^* R^{1,3}\times C^2_L\times C^2_R$, where $T^* R^{1,3}=R^{1,3}\times R^{1,3}$ is the space of coordinates and momenta, and $C^2_L$ and $C^2_R$ are the spaces of representation of the Lorentz group of type $(\frac12 , 0)$ and $(0, \frac12)$. Passing from relativistic mechanics with a Lorentz-invariant Hamiltonian function $H$ on the phase space $X$ to quantum mechanics with a Hamiltonian operator $\hat H$, we introduce two complex conjugate line bundles $L_C^+$ and $L_C^-$ over $X$. Quantum particles are introduced as sections $\Psi_+$ of the bundle $L_C^+$ holomorphic along the space $C^2_L\times C^2_R$, and antiparticles are sections $\Psi_-^{}$ of the bundle $L_C^-$ antiholomorphic along the internal spin space $C^2_L\times C^2_R$. The wave functions $\Psi_\pm$ are characterized by conserved charges $q_{\sf{v}}=\pm 1$ associated with the structure group U(1)$_{\sf{v}}$ of the bundles $L_C^\pm$. Wave functions $\Psi_\pm$ are governed by relativistic analogue of the Schr\"odinger equation. We show how fields with spin $s=0$ (Klein-Gordon), spin $s=\frac12$ (Dirac) and spin $s=1$ (Proca fields) arise from these equations in the zeroth, first, and second order expansions of the functions $\Psi_\pm^{}$ in the coordinates of the spin space $C^2_L\times C^2_R$. The Klein-Gordon, Dirac and Proca equations for these fields follow from the Schr\"odinger equation on the extended phase space $T^* R^{1,3}\times C^2_L\times C^2_R$. Using these results, we also introduce equations describing first quantized photons. We show that taking into account the charges $q_{\sf{v}}=\pm 1$ of the fields $\Psi_\pm$ changes the definitions of the inner products and currents, which eliminates negative energies and negative probabilities from relativistic quantum mechanics.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 12:47:56 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 08:02:54 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Popov",
"Alexander D.",
""
]
] |
2508.21770
|
Qiyue Sun
|
Qiyue Sun, Qiming Huang, Yang Yang, Hongjun Wang, Jianbo Jiao
|
What Can We Learn from Harry Potter? An Exploratory Study of Visual Representation Learning from Atypical Videos
|
Accepted to BMVC 2025
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Humans usually show exceptional generalisation and discovery ability in the open world, when being shown uncommon new concepts. Whereas most existing studies in the literature focus on common typical data from closed sets, open-world novel discovery is under-explored in videos. In this paper, we are interested in asking: What if atypical unusual videos are exposed in the learning process? To this end, we collect a new video dataset consisting of various types of unusual atypical data (e.g., sci-fi, animation, etc.). To study how such atypical data may benefit open-world learning, we feed them into the model training process for representation learning. Focusing on three key tasks in open-world learning: out-of-distribution (OOD) detection, novel category discovery (NCD), and zero-shot action recognition (ZSAR), we found that even straightforward learning approaches with atypical data consistently improve performance across various settings. Furthermore, we found that increasing the categorical diversity of the atypical samples further boosts OOD detection performance. Additionally, in the NCD task, using a smaller yet more semantically diverse set of atypical samples leads to better performance compared to using a larger but more typical dataset. In the ZSAR setting, the semantic diversity of atypical videos helps the model generalise better to unseen action classes. These observations in our extensive experimental evaluations reveal the benefits of atypical videos for visual representation learning in the open world, together with the newly proposed dataset, encouraging further studies in this direction. The project page is at: https://julysun98.github.io/atypical_dataset.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 16:43:19 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 12:51:08 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Sun",
"Qiyue",
""
],
[
"Huang",
"Qiming",
""
],
[
"Yang",
"Yang",
""
],
[
"Wang",
"Hongjun",
""
],
[
"Jiao",
"Jianbo",
""
]
] |
2509.00023
|
Ralf Schwamborn
|
Ralf Schwamborn
|
Towards a compleat theory of ecosystem size spectra
| null | null | null | null |
q-bio.PE
|
http://creativecommons.org/licenses/by/4.0/
|
The regularity of ecosystem size spectra is one of the most intriguing and relevant phenomena on our planet. Aquatic size spectra generally show a log-linearly downtrending shape, following a power-law distribution. A constant log-linear slope has been reported for many marine pelagic ecosystems, often being approximately b = -1. Conversely, there are variable trophic-level-biomass relationships (trophic pyramids). The contrasting observations of a constant biomass (M) spectrum slope and highly variable biomass pyramids may be defined as the constant size spectra-variable trophic dynamics paradox. Here, a 'mass-specific predator-prey-efficiency theory of size spectra' (PETS) is presented and discussed. A thorough analysis of available data, literature and models result in the conclusion that most pelagic marine ecosystems are controlled by trophic processes such as resource-limit stress (bottom-up control) and top-down regulation, with a key role of the maximum carrying capacity of large-sized organisms. This has relevant consequences for the prediction and interpretation of size spectra and the context of fisheries, whaling, and the introduction of exotic predators (e.g., lionfish). The complete size spectrum obtained in situ, including living organisms and non-living particles (e.g., for data from LOPC and UVP) is discussed. This paper is intended as a plea for the integration of modeling approaches, to understand and integrate data and processes across communities including bacteria, phytoplankton, fish and mammals, considering the effect of non-organismic particles.
|
[
{
"version": "v1",
"created": "Mon, 18 Aug 2025 18:30:20 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 03:50:10 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Schwamborn",
"Ralf",
""
]
] |
2509.00054
|
Haimei Pan
|
Haimei Pan, Jiyun Zhang, Qinxi Wei, Xiongnan Jin, Chen Xinkai, Jie Cheng
|
Robotic Fire Risk Detection based on Dynamic Knowledge Graph Reasoning: An LLM-Driven Approach with Graph Chain-of-Thought
|
We have decided to withdraw this paper as the work is still undergoing further refinement. To ensure the clarity of the results, we prefer to make additional improvements before resubmission. We appreciate the readers' understanding
| null | null | null |
cs.RO cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Fire is a highly destructive disaster, but effective prevention can significantly reduce its likelihood of occurrence. When it happens, deploying emergency robots in fire-risk scenarios can help minimize the danger to human responders. However, current research on pre-disaster warnings and disaster-time rescue still faces significant challenges due to incomplete perception, inadequate fire situational awareness, and delayed response. To enhance intelligent perception and response planning for robots in fire scenarios, we first construct a knowledge graph (KG) by leveraging large language models (LLMs) to integrate fire domain knowledge derived from fire prevention guidelines and fire rescue task information from robotic emergency response documents. We then propose a new framework called Insights-on-Graph (IOG), which integrates the structured fire information of KG and Large Multimodal Models (LMMs). The framework generates perception-driven risk graphs from real-time scene imagery to enable early fire risk detection and provide interpretable emergency responses for task module and robot component configuration based on the evolving risk situation. Extensive simulations and real-world experiments show that IOG has good applicability and practical application value in fire risk detection and rescue decision-making.
|
[
{
"version": "v1",
"created": "Mon, 25 Aug 2025 07:10:16 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 05:25:44 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Pan",
"Haimei",
""
],
[
"Zhang",
"Jiyun",
""
],
[
"Wei",
"Qinxi",
""
],
[
"Jin",
"Xiongnan",
""
],
[
"Xinkai",
"Chen",
""
],
[
"Cheng",
"Jie",
""
]
] |
2509.00056
|
Luu Tu Nguyen
|
Luu Tu Nguyen, Vu Tram Anh Khuong, Thanh Ha Le, Thi Duyen Ngo
|
MESTI-MEGANet: Micro-expression Spatio-Temporal Image and Micro-expression Gradient Attention Networks for Micro-expression Recognition
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Micro-expression recognition (MER) is a challenging task due to the subtle and fleeting nature of micro-expressions. Traditional input modalities, such as Apex Frame, Optical Flow, and Dynamic Image, often fail to adequately capture these brief facial movements, resulting in suboptimal performance. In this study, we introduce the Micro-expression Spatio-Temporal Image (MESTI), a novel dynamic input modality that transforms a video sequence into a single image while preserving the essential characteristics of micro-movements. Additionally, we present the Micro-expression Gradient Attention Network (MEGANet), which incorporates a novel Gradient Attention block to enhance the extraction of fine-grained motion features from micro-expressions. By combining MESTI and MEGANet, we aim to establish a more effective approach to MER. Extensive experiments were conducted to evaluate the effectiveness of MESTI, comparing it with existing input modalities across three CNN architectures (VGG19, ResNet50, and EfficientNetB0). Moreover, we demonstrate that replacing the input of previously published MER networks with MESTI leads to consistent performance improvements. The performance of MEGANet, both with MESTI and Dynamic Image, is also evaluated, showing that our proposed network achieves state-of-the-art results on the CASMEII and SAMM datasets. The combination of MEGANet and MESTI achieves the highest accuracy reported to date, setting a new benchmark for micro-expression recognition. These findings underscore the potential of MESTI as a superior input modality and MEGANet as an advanced recognition network, paving the way for more effective MER systems in a variety of applications.
|
[
{
"version": "v1",
"created": "Mon, 25 Aug 2025 09:49:05 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 09:26:23 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Nguyen",
"Luu Tu",
""
],
[
"Khuong",
"Vu Tram Anh",
""
],
[
"Le",
"Thanh Ha",
""
],
[
"Ngo",
"Thi Duyen",
""
]
] |
2509.00134
|
Cody FitzGerald
|
Cody E. FitzGerald, James P. Keener
|
The effect of predation on the dynamics of Chronic Wasting Disease in deer
| null | null | null | null |
q-bio.PE math.DS
|
http://creativecommons.org/licenses/by/4.0/
|
Chronic Wasting Disease (CWD) is a neurological disease impacting deer, elk, moose, and other cervid populations and is caused by a misfolded protein known as a prion. CWD is difficult to control due to the persistence of prions in the environment. Prions can remain infectious for more than a decade and have been found in soil as well as other environmental vectors, such as ticks and plants. Here, we provide a bifurcation analysis of a mathematical model of CWD spread in a cervid population, and use a modification of the Gillespie algorithm to explore if wolves can be used as an ecological control strategy to limit the spread of the disease in several relevant scenarios. We then analytically compute the probability that the disease spreads given one infected member enters a fully healthy population and the probability of elimination, given a fully susceptible population and remaining prions in the environment. From our analysis, we conclude that wolves can be used as an effective control strategy to limit the spread of CWD in cervid populations, and hunting or other means of lowering the susceptible population are beneficial to controlling the spread of CWD, although it is important to note that inferring biologically relevant parameters from the existing data is an ongoing challenge for this system.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 14:51:08 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 02:16:28 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"FitzGerald",
"Cody E.",
""
],
[
"Keener",
"James P.",
""
]
] |
2509.00156
|
Jos\'e De Jes\'us Padua Arg\"uelles
|
Laurent Freidel, Jos\'e Padua-Arg\"uelles, Susanne Schander, Marc Schiffer
|
Non-Perturbative $S$-matrix Renormalization
|
6 pages, one table, one figure, Display of equation 6 fixed
| null | null | null |
hep-th
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We propose a renormalization group flow equation for a functional that generates $S$-matrix elements and which captures similarities to the well-known Wetterich and Polchinski equations. While the latter ones respectively involve the effective action and Schwinger functional, which are genuine off-shell objects, the presented flow equation has the advantage of working more directly with observables, i.e. scattering amplitudes. Compared to the Wetterich equation, our flow equation also greatly simplifies the notion of going on-shell, in the sense of satisfying the quantum equations of motion. In addition, unlike the Wetterich equation, it is polynomial and does not require a Hessian inversion. The approach is a promising direction for non-perturbative quantum field theories, allowing one to work more directly with scattering amplitudes.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 18:00:03 GMT"
},
{
"version": "v2",
"created": "Fri, 05 Sep 2025 22:03:22 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Freidel",
"Laurent",
""
],
[
"Padua-Argüelles",
"José",
""
],
[
"Schander",
"Susanne",
""
],
[
"Schiffer",
"Marc",
""
]
] |
2509.00179
|
Daniel Ablin
|
Daniel Ablin and Alon Cohen
|
Playing Markov Games Without Observing Payoffs
| null | null | null | null |
cs.GT cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Optimization under uncertainty is a fundamental problem in learning and decision-making, particularly in multi-agent systems. Previously, Feldman, Kalai, and Tennenholtz [2010] demonstrated the ability to efficiently compete in repeated symmetric two-player matrix games without observing payoffs, as long as the opponents actions are observed. In this paper, we introduce and formalize a new class of zero-sum symmetric Markov games, which extends the notion of symmetry from matrix games to the Markovian setting. We show that even without observing payoffs, a player who knows the transition dynamics and observes only the opponents sequence of actions can still compete against an adversary who may have complete knowledge of the game. We formalize three distinct notions of symmetry in this setting and show that, under these conditions, the learning problem can be reduced to an instance of online learning, enabling the player to asymptotically match the return of the opponent despite lacking payoff observations. Our algorithms apply to both matrix and Markov games, and run in polynomial time with respect to the size of the game and the number of episodes. Our work broadens the class of games in which robust learning is possible under severe informational disadvantage and deepens the connection between online learning and adversarial game theory.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 18:28:42 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 07:52:00 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Ablin",
"Daniel",
""
],
[
"Cohen",
"Alon",
""
]
] |
2509.00193
|
Lise-Marie Imbert-G\'erard
|
Lise-Marie Imbert-G\'erard
|
A quasi-Trefftz space for a second order time-harmonic Maxwell's equation
| null | null | null | null |
math.NA cs.NA
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Quasi-Trefftz methods are a family of Discontinuous Galerkin methods relying on equation-dependent function spaces. This work is the first study of the notion of local Taylor-based polynomial quasi-Trefftz space for a system of Partial Differential Equations (PDEs). These discrete spaces are introduced here for electro-magnetic wave propagation in inhomogeneous media, governed by a second order formulation of Maxwell's equation with variable coefficients. Thanks to an adequate Helmholtz decomposition for spaces of homogeneous polynomial vector fields, the outcome is the explicit dimension of the proposed quasi-Trefftz space as well as a procedure to construct quasi-Trefftz functions.
|
[
{
"version": "v1",
"created": "Fri, 29 Aug 2025 18:58:41 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 01:04:31 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Imbert-Gérard",
"Lise-Marie",
""
]
] |
2509.00340
|
Abdullah Zayat
|
Abdullah Zayat, Omran Abbas, Loic Markley, and Anas Chaaban
|
Deep Complex-valued Neural-Network Modeling and Optimization of Stacked Intelligent Surfaces
| null | null |
10.1109/MeditCom64437.2025.11104397
| null |
cs.IT math.IT
|
http://creativecommons.org/licenses/by/4.0/
|
We propose a complex-valued neural-network (CV-NN) framework to optimally configure stacked intelligent surfaces (SIS) in next-generation multi-antenna systems. Unlike conventional solutions that separately tune analog metasurface phases or rely strictly on SVD-based orthogonal decompositions, our method models each SIS element as a unit-modulus complex-velued neuron in an end-to-end differentiable pipeline. This approach avoids enforcing channel orthogonality and instead allows for richer wavefront designs that can target a wide range of system objectives, such as maximizing spectral efficiency and minimizing detection errors, all within a single optimization framework. Moreover, by exploiting a fully differentiable neural-network formulation and GPU-based auto-differentiation, our approach can rapidly train SIS configurations for realistic, high-dimensional channels, enabling near-online adaptation. Our framework also naturally accommodates hybrid analog-digital beamforming and recovers classical SVD solutions as a special case. Numerical evaluations under Rician channels demonstrate that CV-NN SIS optimization outperforms state-of-the-art schemes in throughput, error performance, and robustness to channel variation, opening the door to more flexible and powerful wave-domain control for future 6G networks.
|
[
{
"version": "v1",
"created": "Sat, 30 Aug 2025 03:45:11 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Zayat",
"Abdullah",
""
],
[
"Abbas",
"Omran",
""
],
[
"Markley",
"Loic",
""
],
[
"Chaaban",
"Anas",
""
]
] |
2509.00457
|
Salah Eddine Bekhouche Dr.
|
Salah Eddine Bekhouche, Abdellah Zakaria Sellam, Hichem Telli, Cosimo Distante, Abdenour Hadid
|
CVPD at QIAS 2025 Shared Task: An Efficient Encoder-Based Approach for Islamic Inheritance Reasoning
| null | null | null | null |
cs.CL cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Islamic inheritance law (Ilm al-Mawarith) requires precise identification of heirs and calculation of shares, which poses a challenge for AI. In this paper, we present a lightweight framework for solving multiple-choice inheritance questions using a specialised Arabic text encoder and Attentive Relevance Scoring (ARS). The system ranks answer options according to semantic relevance, and enables fast, on-device inference without generative reasoning. We evaluate Arabic encoders (MARBERT, ArabicBERT, AraBERT) and compare them with API-based LLMs (Gemini, DeepSeek) on the QIAS 2025 dataset. While large models achieve an accuracy of up to 87.6%, they require more resources and are context-dependent. Our MARBERT-based approach achieves 69.87% accuracy, presenting a compelling case for efficiency, on-device deployability, and privacy. While this is lower than the 87.6% achieved by the best-performing LLM, our work quantifies a critical trade-off between the peak performance of large models and the practical advantages of smaller, specialized systems in high-stakes domains.
|
[
{
"version": "v1",
"created": "Sat, 30 Aug 2025 11:03:54 GMT"
},
{
"version": "v2",
"created": "Fri, 05 Sep 2025 20:27:56 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Bekhouche",
"Salah Eddine",
""
],
[
"Sellam",
"Abdellah Zakaria",
""
],
[
"Telli",
"Hichem",
""
],
[
"Distante",
"Cosimo",
""
],
[
"Hadid",
"Abdenour",
""
]
] |
2509.00575
|
Himanshu Verma
|
Himanshu Verma, Kirtan Padh, Eva Thelisson
|
Can AI be Auditable?
| null | null | null | null |
cs.CY cs.AI cs.HC
|
http://creativecommons.org/licenses/by/4.0/
|
Auditability is defined as the capacity of AI systems to be independently assessed for compliance with ethical, legal, and technical standards throughout their lifecycle. The chapter explores how auditability is being formalized through emerging regulatory frameworks, such as the EU AI Act, which mandate documentation, risk assessments, and governance structures. It analyzes the diverse challenges facing AI auditability, including technical opacity, inconsistent documentation practices, lack of standardized audit tools and metrics, and conflicting principles within existing responsible AI frameworks. The discussion highlights the need for clear guidelines, harmonized international regulations, and robust socio-technical methodologies to operationalize auditability at scale. The chapter concludes by emphasizing the importance of multi-stakeholder collaboration and auditor empowerment in building an effective AI audit ecosystem. It argues that auditability must be embedded in AI development practices and governance infrastructures to ensure that AI systems are not only functional but also ethically and legally aligned.
|
[
{
"version": "v1",
"created": "Sat, 30 Aug 2025 18:03:20 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 12:26:44 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Verma",
"Himanshu",
""
],
[
"Padh",
"Kirtan",
""
],
[
"Thelisson",
"Eva",
""
]
] |
2509.00651
|
Tin Luu Trung
|
Tin Luu, Binh Nguyen, Man Ngo
|
Missing Data Imputation using Neural Cellular Automata
|
20 pages, 4 figures
| null | null | null |
cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
When working with tabular data, missingness is always one of the most painful problems. Throughout many years, researchers have continuously explored better and better ways to impute missing data. Recently, with the rapid development evolution in machine learning and deep learning, there is a new trend of leveraging generative models to solve the imputation task. While the imputing version of famous models such as Variational Autoencoders or Generative Adversarial Networks were investigated, prior work has overlooked Neural Cellular Automata (NCA), a powerful computational model. In this paper, we propose a novel imputation method that is inspired by NCA. We show that, with some appropriate adaptations, an NCA-based model is able to address the missing data imputation problem. We also provide several experiments to evidence that our model outperforms state-of-the-art methods in terms of imputation error and post-imputation performance.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 01:13:59 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 11:04:54 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Luu",
"Tin",
""
],
[
"Nguyen",
"Binh",
""
],
[
"Ngo",
"Man",
""
]
] |
2509.00656
|
Shu-Hua Yang
|
Shu-Hua Yang, Fridolin Weber
|
XTE J1814-338 as a strange star admixed with bosonic dark matter
|
5 pages, 1 figure
| null | null | null |
astro-ph.HE hep-ph
|
http://creativecommons.org/licenses/by/4.0/
|
We show that the compact star XTE J1814-338 can be explained as a strange star admixed with self-interacting bosonic dark matter (BDM), provided the dark matter fraction exceeds approximately 70\%. This interpretation leads to a robust constraint on the BDM particle mass: $m_\chi \lesssim 307(\lambda/\pi)^{1/4}$ MeV ($\lambda$ is the dimensionless coupling constant of the BDM). The result is independent of formation scenario and microphysical details and is falsifiable by future NICER and LIGO/Virgo observations.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 01:39:17 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 08:38:37 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yang",
"Shu-Hua",
""
],
[
"Weber",
"Fridolin",
""
]
] |
2509.00662
|
Vamsi Shankar Simhadri
|
Vamsi Shankar Simhadri, Yichang Xiong, Habiba Farrukh, Xiaokuan Zhang
|
Virtual Reality, Real Problems: A Longitudinal Security Analysis of VR Firmware
|
17 pages, 25 figures, 6 tables, To appear on ACM CCS 2025
| null | null | null |
cs.CR
|
http://creativecommons.org/licenses/by/4.0/
|
Virtual Reality (VR) technology is rapidly growing in recent years. VR devices such as Meta Quest 3 utilize numerous sensors to collect users' data to provide an immersive experience. Due to the extensive data collection and the immersive nature, the security of VR devices is paramount. Leading VR devices often adopt and customize Android systems, which makes them susceptible to both Android-based vulnerabilities and new issues introduced by VR-specific customizations (e.g., system services to support continuous head and hand tracking). While prior work has extensively examined the security properties of the Android software stack, how these security properties hold for VR systems remains unexplored. In this paper, we present the first comprehensive security analysis of VR firmware. We collect over 300 versions of VR firmware from two major vendors, Quest and Pico, and perform a longitudinal analysis across the kernel layer, the system binary and library layer, and the application layer. We have identified several security issues in these VR firmware, including missing kernel-level security features, insufficient binary hardening, inconsistent permission enforcement, and inadequate SELinux policy enforcement. Based on our findings, we synthesize recommendations for VR vendors to improve security and trust for VR devices. This paper will act as an important security resource for VR developers, users, and vendors, and will also direct future advancements in secure VR ecosystem.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 02:16:56 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 17:10:40 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Simhadri",
"Vamsi Shankar",
""
],
[
"Xiong",
"Yichang",
""
],
[
"Farrukh",
"Habiba",
""
],
[
"Zhang",
"Xiaokuan",
""
]
] |
2509.00665
|
Weilong Yan
|
Weilong Yan, Xin Zhang, Robby T. Tan
|
ER-LoRA: Effective-Rank Guided Adaptation for Weather-Generalized Depth Estimation
| null | null | null | null |
cs.CV cs.RO
|
http://creativecommons.org/licenses/by/4.0/
|
Monocular depth estimation under adverse weather conditions (e.g.\ rain, fog, snow, and nighttime) remains highly challenging due to the lack of reliable ground truth and the difficulty of learning from unlabeled real-world data. Existing methods often rely on synthetic adverse data with pseudo-labels, which suffer from domain gaps, or employ self-supervised learning, which violates photometric assumptions in adverse scenarios. In this work, we propose to achieve weather-generalized depth estimation by Parameter-Efficient Fine-Tuning (PEFT) of Vision Foundation Models (VFMs), using only a small amount of high-visibility (normal) data. While PEFT has shown strong performance in semantic tasks such as segmentation, it remains underexplored for geometry -- centric tasks like depth estimation -- especially in terms of balancing effective adaptation with the preservation of pretrained knowledge. To this end, we introduce the Selecting-Tuning-Maintaining (STM) strategy, which structurally decomposes the pretrained weights of VFMs based on two kinds of effective ranks (entropy-rank and stable-rank). In the tuning phase, we adaptively select the proper rank number as well as the task-aware singular directions for initialization, based on the entropy-rank and full-tuned weight; while in the maintaining stage, we enforce a principal direction regularization based on the stable-rank. This design guarantees flexible task adaptation while preserving the strong generalization capability of the pretrained VFM. Extensive experiments on four real-world benchmarks across diverse weather conditions demonstrate that STM not only outperforms existing PEFT methods and full fine-tuning but also surpasses methods trained with adverse synthetic data, and even the depth foundation model
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 02:24:00 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 16:03:21 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yan",
"Weilong",
""
],
[
"Zhang",
"Xin",
""
],
[
"Tan",
"Robby T.",
""
]
] |
2509.00686
|
Thomas Curtright
|
Thomas Curtright and Galib Hoq
|
Analytic Scalar Field Theory
|
Footnote and reference added in v2
| null | null | null |
hep-th math-ph math.MP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We discuss scalar field theories with potentials V({\phi})=\k{appa}({\phi}^2)^{{\nu}} for generic {\nu}. We conjecture that these models evade triviality theorems for scalar fields.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 03:58:34 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 19:07:10 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Curtright",
"Thomas",
""
],
[
"Hoq",
"Galib",
""
]
] |
2509.00787
|
Ganxi Xu
|
Ganxi Xu, Jinyi Long, Jia Zhang
|
Diffusion-Based Image-to-Brain Signal Generation with Cross-Attention Mechanisms for Visual Prostheses
| null | null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Visual prostheses have shown great potential in restoring vision for blind individuals. However, while researchers have successfully utilized M/EEG signals to evoke visual perceptions during the brain decoding stage of visual prostheses, the complementary process-converting images to M/EEG signals in the brain encoding stage-remains largely unexplored. Thus, we present the first image-to-brain signal (M/EEG) framework based on denoising diffusion probabilistic models enhanced with cross-attention mechanisms. Our framework consists of two key architectural components: a pre-trained CLIP visual encoder that extracts rich semantic representations from input images, and a cross-attention enhanced U-Net diffusion model that learns to reconstruct biologically plausible brain signals through iterative denoising. Unlike conventional generative models that rely on simple concatenation for conditioning, our cross-attention modules enable dynamic interaction between visual features and brain signal representations, facilitating fine-grained alignment during the generation process. Furthermore, we evaluate our framework on two multimodal datasets (THINGS-EEG2 and THINGS-MEG) to demonstrate its effectiveness in generating biologically plausible brain signals. Additionally, we pioneer the visualization of M/EEG topographies across all subjects in both datasets, providing intuitive demonstrations of intra-subject and inter-subject variations in brain signals.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 10:29:58 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 23:57:27 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Xu",
"Ganxi",
""
],
[
"Long",
"Jinyi",
""
],
[
"Zhang",
"Jia",
""
]
] |
2509.00823
|
Akira Terui
|
Takumu Okazaki, Akira Terui, Masahiko Mikawa
|
Inverse Kinematics for a 6-Degree-of-Freedom Robot Manipulator Using Comprehensive Gr\"obner Systems
|
24 pages
| null | null | null |
cs.RO cs.SC math.AC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We propose an effective method for solving the inverse kinematic problem of a specific model of 6-degree-of-freedom (6-DOF) robot manipulator using computer algebra. It is known that when the rotation axes of three consecutive rotational joints of a manipulator intersect at a single point, the inverse kinematics problem can be divided into determining position and orientation. We extend this method to more general manipulators in which the rotational axes of two consecutive joints intersect. This extension broadens the class of 6-DOF manipulators for which the inverse kinematics problem can be solved, and is expected to enable more efficient solutions. The inverse kinematic problem is solved using the Comprehensive Gr\"obner System (CGS) with joint parameters of the robot appearing as parameters in the coefficients to prevent repetitive calculations of the Gr\"obner bases. The effectiveness of the proposed method is shown by experiments.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 12:41:32 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 07:02:03 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Okazaki",
"Takumu",
""
],
[
"Terui",
"Akira",
""
],
[
"Mikawa",
"Masahiko",
""
]
] |
2509.00828
|
Akira Terui
|
Takumu Okazaki, Akira Terui, Masahiko Mikawa
|
An Effective Trajectory Planning and an Optimized Path Planning for a 6-Degree-of-Freedom Robot Manipulator
|
26 pages
| null | null | null |
cs.RO cs.SC math.AC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
An effective method for optimizing path planning for a specific model of a 6-degree-of-freedom (6-DOF) robot manipulator is presented as part of the motion planning of the manipulator using computer algebra. We assume that we are given a path in the form of a set of line segments that the end-effector should follow. We also assume that we have a method to solve the inverse kinematic problem of the manipulator at each via-point of the trajectory. The proposed method consists of three steps. First, we calculate the feasible region of the manipulator under a specific configuration of the end-effector. Next, we aim to find a trajectory on the line segments and a sequence of joint configurations the manipulator should follow to move the end-effector along the specified trajectory. Finally, we find the optimal combination of solutions to the inverse kinematic problem at each via-point along the trajectory by reducing the problem to a shortest-path problem of the graph and applying Dijkstra's algorithm. We show the effectiveness of the proposed method by experiments.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 12:59:46 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 07:08:38 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Okazaki",
"Takumu",
""
],
[
"Terui",
"Akira",
""
],
[
"Mikawa",
"Masahiko",
""
]
] |
2509.00856
|
Alexander Komech
|
A.I. Komech and E.A. Kopylova
|
On dissipation operators of Quantum Optics
|
5 pages, 0 figures
| null | null | null |
quant-ph math-ph math.MP
|
http://creativecommons.org/licenses/by/4.0/
|
We consider dissipation operators used in Quantum Optics for the description of quantum spontaneous emission in the context of damped driven Jaynes-Cummings equations. The equations describe quantised one-mode Maxwell field coupled to a two-level molecule. Our main result is the symmetry and nonpositivity of basic dissipation operator of Quantum Optics.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 14:02:41 GMT"
},
{
"version": "v2",
"created": "Wed, 03 Sep 2025 07:11:43 GMT"
},
{
"version": "v3",
"created": "Sun, 07 Sep 2025 10:21:34 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Komech",
"A. I.",
""
],
[
"Kopylova",
"E. A.",
""
]
] |
2509.00859
|
Jiacheng Jiang
|
Jiacheng Jiang, Yuan Meng, Chen Tang, Han Yu, Qun Li, Zhi Wang, Wenwu Zhu
|
Quantization Meets OOD: Generalizable Quantization-aware Training from a Flatness Perspective
| null |
Proc. of the 33rd ACM International Conference on Multimedia (MM '25), Dublin, Ireland, October 2025
|
10.1145/3746027.3755856
| null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Current quantization-aware training (QAT) methods primarily focus on enhancing the performance of quantized models on in-distribution (I.D) data, while overlooking the potential performance degradation on out-of-distribution (OOD) data. In this paper, we first substantiate this problem through rigorous experiment, showing that QAT can lead to a significant OOD generalization performance degradation. Further, we find the contradiction between the perspective that flatness of loss landscape gives rise to superior OOD generalization and the phenomenon that QAT lead to a sharp loss landscape, can cause the above problem. Therefore, we propose a flatness-oriented QAT method, FQAT, to achieve generalizable QAT. Specifically, i) FQAT introduces a layer-wise freezing mechanism to mitigate the gradient conflict issue between dual optimization objectives (i.e., vanilla QAT and flatness). ii) FQAT proposes an disorder-guided adaptive freezing algorithm to dynamically determines which layers to freeze at each training step, effectively addressing the challenges caused by interference between layers. A gradient disorder metric is designed to help the algorithm identify unstable layers during training. Extensive experiments on influential OOD benchmark demonstrate the superiority of our method over state-of-the-art baselines under both I.D and OOD image classification tasks.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 14:05:14 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Jiang",
"Jiacheng",
""
],
[
"Meng",
"Yuan",
""
],
[
"Tang",
"Chen",
""
],
[
"Yu",
"Han",
""
],
[
"Li",
"Qun",
""
],
[
"Wang",
"Zhi",
""
],
[
"Zhu",
"Wenwu",
""
]
] |
2509.00882
|
Xiang Li
|
Xiang Li and Yueci Su and Jiahao Liu and Zhiwei Lin and Yuebing Hou and Peiming Gao and Yuanchao Zhang
|
VulSolver: Vulnerability Detection via LLM-Driven Constraint Solving
| null | null | null | null |
cs.CR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Traditional vulnerability detection methods rely heavily on predefined rule matching, which often fails to capture vulnerabilities accurately. With the rise of large language models (LLMs), leveraging their ability to understand code semantics has emerged as a promising direction for achieving more accurate and efficient vulnerability detection. However, current LLM-based approaches face significant challenges: instability in model outputs, limitations in context length, and hallucination. As a result, many existing solutions either use LLMs merely to enrich predefined rule sets, thereby keeping the detection process fundamentally rule-based, or over-rely on them, leading to poor robustness. To address these challenges, we propose a constraint-solving approach powered by LLMs named VULSOLVER. By modeling vulnerability detection as a constraint-solving problem, and by integrating static application security testing (SAST) with the semantic reasoning capabilities of LLMs, our method enables the LLM to act like a professional human security expert. We assess VULSOLVER on the OWASP Benchmark (1,023 labeled samples), achieving 96.29% accuracy, 96.55% F1-score, and 100% recall. Applied to popular GitHub repositories, VULSOLVER also identified 15 previously unknown high-severity vulnerabilities (CVSS 7.5-9.8), demonstrating its effectiveness in real-world security analysis.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 14:49:48 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 08:20:32 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Li",
"Xiang",
""
],
[
"Su",
"Yueci",
""
],
[
"Liu",
"Jiahao",
""
],
[
"Lin",
"Zhiwei",
""
],
[
"Hou",
"Yuebing",
""
],
[
"Gao",
"Peiming",
""
],
[
"Zhang",
"Yuanchao",
""
]
] |
2509.00948
|
Denghang Hu
|
Denghang Hu, Taolue Chen, Philipp R\"ummer, Fu Song, Zhilin Wu
|
Decision Procedure for A Theory of String Sequences
|
21 pages, 2 tables, APLAS 2025
| null | null | null |
cs.PL cs.FL
|
http://creativecommons.org/licenses/by/4.0/
|
The theory of sequences, supported by many SMT solvers, can model program data types including bounded arrays and lists. Sequences are parameterized by the element data type and provide operations such as accessing elements, concatenation, forming sub-sequences and updating elements. Strings and sequences are intimately related; many operations, e.g., matching a string according to a regular expression, splitting strings, or joining strings in a sequence, are frequently used in string-manipulating programs. Nevertheless, these operations are typically not directly supported by existing SMT solvers, which instead only consider the generic theory of sequences. In this paper, we propose a theory of string sequences and study its satisfiability. We show that, while it is undecidable in general, the decidability can be recovered by restricting to the straight-line fragment. This is shown by encoding each string sequence as a string, and each string sequence operation as a corresponding string operation. We provide pre-image computation for the resulting string operations with respect to automata, effectively casting it into the generic OSTRICH string constraint solving framework. We implement the new decision procedure as a tool $\ostrichseq$, and carry out experiments on benchmark constraints generated from real-world JavaScript programs, hand-crafted templates and unit tests. The experiments confirm the efficacy of our approach.
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 17:57:01 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 04:40:03 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Hu",
"Denghang",
""
],
[
"Chen",
"Taolue",
""
],
[
"Rümmer",
"Philipp",
""
],
[
"Song",
"Fu",
""
],
[
"Wu",
"Zhilin",
""
]
] |
2509.01025
|
Jaeyeon Kim
|
Jaeyeon Kim, Lee Cheuk-Kit, Carles Domingo-Enrich, Yilun Du, Sham Kakade, Timothy Ngotiaoco, Sitan Chen, Michael Albergo
|
Any-Order Flexible Length Masked Diffusion
|
Preprint
| null | null | null |
cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Masked diffusion models (MDMs) have recently emerged as a promising alternative to autoregressive models over discrete domains. MDMs generate sequences in an any-order, parallel fashion, enabling fast inference and strong performance on non-causal tasks. However, a crucial limitation is that they do not support token insertions and are thus limited to fixed-length generations. To this end, we introduce Flexible Masked Diffusion Models (FlexMDMs), a discrete diffusion paradigm that simultaneously can model sequences of flexible length while provably retaining MDMs' flexibility of any-order inference. Grounded in an extension of the stochastic interpolant framework, FlexMDMs generate sequences by inserting mask tokens and unmasking them. Empirically, we show that FlexMDMs match MDMs in perplexity while modeling length statistics with much higher fidelity. On a synthetic maze planning task, they achieve $\approx 60 \%$ higher success rate than MDM baselines. Finally, we show pretrained MDMs can easily be retrofitted into FlexMDMs: on 16 H100s, it takes only three days to fine-tune LLaDA-8B into a FlexMDM, achieving superior performance on math (GSM8K, $58\% \to 67\%$) and code infilling performance ($52\% \to 65\%$).
|
[
{
"version": "v1",
"created": "Sun, 31 Aug 2025 23:34:53 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 22:48:13 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Kim",
"Jaeyeon",
""
],
[
"Cheuk-Kit",
"Lee",
""
],
[
"Domingo-Enrich",
"Carles",
""
],
[
"Du",
"Yilun",
""
],
[
"Kakade",
"Sham",
""
],
[
"Ngotiaoco",
"Timothy",
""
],
[
"Chen",
"Sitan",
""
],
[
"Albergo",
"Michael",
""
]
] |
2509.01083
|
Eunjoo Jeon
|
Mingyu Yang, Jae-Young Choi, Kihyo Moon, Minsung Jang, and Eunjoo Jeon
|
DSDE: Dynamic Speculative Decoding with KLD Stability for Real-World Serving
|
10 pages, 9 figures. Preprint submitted to IEEE BigData 2025
| null | null | null |
cs.DC cs.AI cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Speculative decoding accelerates large language model inference, but its reliance on a fixed speculation length is suboptimal in large-batch serving environments with diverse requests. This paper explores a new direction for dynamic adaptation by investigating a novel class of post-hoc, diagnostic signals. We propose Dynamic Speculative Decoding Engine (DSDE), a training-free framework built on two primary components: (1) a predictive signal based on the variance of the Kullback-Leibler (KLD) divergence, which diagnoses the generation's regional stability, and (2) an adaptive speculation length cap to mitigate the straggler problem in per-sequence decoding. Experiments demonstrate the potential of using KLD-based stability signals for dynamic adaptation. An algorithm guided by these signals achieves end-to-end latency competitive with leading baselines and exhibits superior robustness across diverse workloads. This robustness is particularly valuable in challenging low-acceptance-rate regimes, where the proposed signal maintains its diagnostic utility. Collectively, these findings validate post-hoc signals as a valuable component for building more robust and intelligent LLM inference systems, and highlight a promising direction for future research on dynamic speculation length adaptation.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 03:13:50 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 03:27:39 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yang",
"Mingyu",
""
],
[
"Choi",
"Jae-Young",
""
],
[
"Moon",
"Kihyo",
""
],
[
"Jang",
"Minsung",
""
],
[
"Jeon",
"Eunjoo",
""
]
] |
2509.01116
|
Yakun Xi
|
Chuanwei Gao and Shukun Wu and Yakun Xi
|
Sharp microlocal Kakeya--Nikodym estimates for H\"ormander operators and spectral projectors
|
33 pages, minor edits
| null | null | null |
math.CA math.AP math.SP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We establish sharp microlocal Kakeya--Nikodym estimates for H\"ormander operators with positive-definite Carleson--Sj\"olin phases and for spectral projectors on smooth, compact Riemannian manifolds. As an application, we obtain sharp $L^q\to L^p$ estimates for the aforementioned H\"ormander operators in odd dimensions, thereby completing the analysis in the odd-dimensional case. Further applications include $L^q\to L^p$ estimates for the Fourier extension operator, $L^p$ estimates for the Bochner--Riesz operator, microlocal Kakeya--Nikodym estimates for Laplace eigenfunctions, and $L^p$ estimates for Hecke--Maass forms on compact $3$-dimensional arithmetic hyperbolic manifolds.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 04:22:34 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 13:52:27 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Gao",
"Chuanwei",
""
],
[
"Wu",
"Shukun",
""
],
[
"Xi",
"Yakun",
""
]
] |
2509.01128
|
Hortense Le Ferrand
|
Aminul Islam, Mukta Bansal, Lena Felix Stephanie, Poernomo Gunawan, Pui Tze Sian, Sabrina Luk, Eunice Tan, Hortense Le Ferrand
|
Assessing prompting frameworks for enhancing literature reviews among university students using ChatGPT
|
24 pages, 6 figures
| null | null | null |
cs.CY
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
Writing literature reviews is a common component of university curricula, yet it often poses challenges for students. Since generative artificial intelligence (GenAI) tools have been made publicly accessible, students have been employing them for their academic writing tasks. However, there is limited evidence of structured training on how to effectively use these GenAI tools to support students in writing literature reviews. In this study, we explore how university students use one of the most popular GenAI tools, ChatGPT, to write literature reviews and how prompting frameworks can enhance their output. To this aim, prompts and literature reviews written by a group of university students were collected before and after they had been introduced to three prompting frameworks, namely CO-STAR, POSE, and Sandwich. The results indicate that after being exposed to these prompting frameworks, the students demonstrated improved prompting behaviour, resulting in more effective prompts and higher quality literature reviews. However, it was also found that the students did not fully utilise all the elements in the prompting frameworks, and aspects such as originality, critical analysis, and depth in their reviews remain areas for improvement. The study, therefore, raises important questions about the significance of utilising prompting frameworks in their entirety to maximise the quality of outcomes, as well as the extent of prior writing experience students should have before leveraging GenAI in the process of writing literature reviews. These findings are of interest for educators considering the integration of GenAI into academic writing tasks such as literature reviews or evaluating whether to permit students to use these tools.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 04:47:34 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 03:39:46 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Islam",
"Aminul",
""
],
[
"Bansal",
"Mukta",
""
],
[
"Stephanie",
"Lena Felix",
""
],
[
"Gunawan",
"Poernomo",
""
],
[
"Sian",
"Pui Tze",
""
],
[
"Luk",
"Sabrina",
""
],
[
"Tan",
"Eunice",
""
],
[
"Ferrand",
"Hortense Le",
""
]
] |
2509.01190
|
Sajjad Kachuee
|
Sajjad Kachuee, Mohammad Sharifkhani
|
Efficient Large Language Models with Zero-Shot Adjustable Acceleration
| null | null | null | null |
cs.CL
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Using Large Language Models (LLMs) in real-world applications presents significant challenges, particularly in balancing computational efficiency with model performance. Optimizing acceleration after fine-tuning and during inference is critical for building efficient architectures. This paper introduces Zero-Shot Adjustable Acceleration, a novel training and inference method that dynamically adjusts hardware utilization during inference without requiring additional fine-tuning. The proposed approach is applied to recent LLMs and evaluated across multiple classification and text generation tasks. Experimental results demonstrate that the method supports a wide range of zero-shot acceleration and achieves up to 11x speedup compared to the baseline.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 07:15:25 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 17:46:00 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Kachuee",
"Sajjad",
""
],
[
"Sharifkhani",
"Mohammad",
""
]
] |
2509.01217
|
Lasse Hansen
|
Lasse Hansen, Wiebke Heyer, Christoph Gro{\ss}br\"ohmer, Frederic Madesta, Thilo Sentker, Wang Jiazheng, Yuxi Zhang, Hang Zhang, Min Liu, Junyi Wang, Xi Zhu, Yuhua Li, Liwen Wang, Daniil Morozov, Nazim Haouchine, Joel Honkamaa, Pekka Marttinen, Yichao Zhou, Zuopeng Tan, Zhuoyuan Wang, Yi Wang, Hongchao Zhou, Shunbo Hu, Yi Zhang, Qian Tao, Lukas F\"orner, Thomas Wendler, Bailiang Jian, Christian Wachinger, Jin Kim, Dan Ruan, Marek Wodzinski, Henning M\"uller, Tony C.W. Mok, Xi Jia, Jinming Duan, Mikael Brudfors, Seyed-Ahmad Ahmadi, Yunzheng Zhu, William Hsu, Tina Kapur, William M. Wells, Alexandra Golby, Aaron Carass, Harrison Bai, Yihao Liu, Perrine Paul-Gilloteaux, Joakim Lindblad, Nata\v{s}a Sladoje, Andreas Walter, Junyu Chen, Reuben Dorent, Alessa Hering, Mattias P. Heinrich
|
Learn2Reg 2024: New Benchmark Datasets Driving Progress on New Challenges
|
submitted to MELBA Journal v2: added Jinming Duan to author list
| null | null | null |
eess.IV cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
Medical image registration is critical for clinical applications, and fair benchmarking of different methods is essential for monitoring ongoing progress. To date, the Learn2Reg 2020-2023 challenges have released several complementary datasets and established metrics for evaluations. However, these editions did not capture all aspects of the registration problem, particularly in terms of modality diversity and task complexity. To address these limitations, the 2024 edition introduces three new tasks, including large-scale multi-modal registration and unsupervised inter-subject brain registration, as well as the first microscopy-focused benchmark within Learn2Reg. The new datasets also inspired new method developments, including invertibility constraints, pyramid features, keypoints alignment and instance optimisation.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 08:03:07 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 11:50:29 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Hansen",
"Lasse",
""
],
[
"Heyer",
"Wiebke",
""
],
[
"Großbröhmer",
"Christoph",
""
],
[
"Madesta",
"Frederic",
""
],
[
"Sentker",
"Thilo",
""
],
[
"Jiazheng",
"Wang",
""
],
[
"Zhang",
"Yuxi",
""
],
[
"Zhang",
"Hang",
""
],
[
"Liu",
"Min",
""
],
[
"Wang",
"Junyi",
""
],
[
"Zhu",
"Xi",
""
],
[
"Li",
"Yuhua",
""
],
[
"Wang",
"Liwen",
""
],
[
"Morozov",
"Daniil",
""
],
[
"Haouchine",
"Nazim",
""
],
[
"Honkamaa",
"Joel",
""
],
[
"Marttinen",
"Pekka",
""
],
[
"Zhou",
"Yichao",
""
],
[
"Tan",
"Zuopeng",
""
],
[
"Wang",
"Zhuoyuan",
""
],
[
"Wang",
"Yi",
""
],
[
"Zhou",
"Hongchao",
""
],
[
"Hu",
"Shunbo",
""
],
[
"Zhang",
"Yi",
""
],
[
"Tao",
"Qian",
""
],
[
"Förner",
"Lukas",
""
],
[
"Wendler",
"Thomas",
""
],
[
"Jian",
"Bailiang",
""
],
[
"Wachinger",
"Christian",
""
],
[
"Kim",
"Jin",
""
],
[
"Ruan",
"Dan",
""
],
[
"Wodzinski",
"Marek",
""
],
[
"Müller",
"Henning",
""
],
[
"Mok",
"Tony C. W.",
""
],
[
"Jia",
"Xi",
""
],
[
"Duan",
"Jinming",
""
],
[
"Brudfors",
"Mikael",
""
],
[
"Ahmadi",
"Seyed-Ahmad",
""
],
[
"Zhu",
"Yunzheng",
""
],
[
"Hsu",
"William",
""
],
[
"Kapur",
"Tina",
""
],
[
"Wells",
"William M.",
""
],
[
"Golby",
"Alexandra",
""
],
[
"Carass",
"Aaron",
""
],
[
"Bai",
"Harrison",
""
],
[
"Liu",
"Yihao",
""
],
[
"Paul-Gilloteaux",
"Perrine",
""
],
[
"Lindblad",
"Joakim",
""
],
[
"Sladoje",
"Nataša",
""
],
[
"Walter",
"Andreas",
""
],
[
"Chen",
"Junyu",
""
],
[
"Dorent",
"Reuben",
""
],
[
"Hering",
"Alessa",
""
],
[
"Heinrich",
"Mattias P.",
""
]
] |
2509.01224
|
Zhenghao Zhong
|
Kwok Ho Wan and Zhenghao Zhong
|
Cutting stabiliser decompositions of magic state cultivation with ZX-calculus
| null | null | null | null |
quant-ph
|
http://creativecommons.org/licenses/by/4.0/
|
We apply the cutting stabiliser decomposition techniques [arXiv:2403.10964] to the quantum states generated from magic state cultivation [arXiv:2409.17595], post-selected upon all $+1$ measured values for simplicity. The resultant states to the $d=3$ and $d=5$ variant magic state cultivation circuits can be expressed as a sum of $4$ and $8$ Clifford ZX-digrams respectively. Modifications to existing ZX-calculus stabiliser decomposition methods may enable better simulation of non-Clifford circuits containing a moderate number of $T$ gates in the context of quantum error correction.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 08:11:42 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 15:03:02 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Wan",
"Kwok Ho",
""
],
[
"Zhong",
"Zhenghao",
""
]
] |
2509.01313
|
Zhao Tian
|
Zhao Tian and Junjie Chen
|
Aligning Requirement for Large Language Model's Code Generation
|
Accepted by ICSE 2026
| null | null | null |
cs.SE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Code generation refers to the automatic generation of source code based on a given programming specification, which has garnered significant attention particularly with the advancement of large language models (LLMs). However, due to the inherent complexity of real-world problems, the LLM-generated code often fails to fully align with the provided specification. While state-of-the-art agent-based techniques have been proposed to enhance LLM code generation, they overlook the critical issue of specification perception, resulting in persistent misalignment issues. Given that accurate perception of programming specifications serves as the foundation of the LLM-based code generation paradigm, ensuring specification alignment is particularly crucial. In this work, we draw on software requirements engineering to propose Specine, a novel specification alignment technique for LLM code generation. Its key idea is to identify misaligned input specifications, lift LLM-perceived specifications, and align them to enhance the code generation performance of LLMs. Our comprehensive experiments on four state-of-the-art LLMs across five challenging competitive benchmarks by comparing with ten state-of-the-art baselines, demonstrate the effectiveness of Specine. For example, Specine outperforms the most effective baseline, achieving an average improvement of 29.60% across all subjects in terms of Pass@1.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 09:56:13 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 01:50:34 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Tian",
"Zhao",
""
],
[
"Chen",
"Junjie",
""
]
] |
2509.01350
|
Yunqing Liu
|
Yunqing Liu, Nan Zhang, Zhiming Tan
|
Error Notebook-Guided, Training-Free Part Retrieval in 3D CAD Assemblies via Vision-Language Models
| null | null | null | null |
cs.AI
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Effective specification-aware part retrieval within complex CAD assemblies is essential for automated design verification and downstream engineering tasks. However, directly using LLMs/VLMs to this task presents some challenges: the input sequences may exceed model token limits, and even after processing, performance remains unsatisfactory. Moreover, fine-tuning LLMs/VLMs requires significant computational resources, and for many high-performing general-use proprietary models (e.g., GPT or Gemini), fine-tuning access is not available. In this paper, we propose a novel part retrieval framework that requires no extra training, but using Error Notebooks + RAG for refined prompt engineering to help improve the existing general model's retrieval performance. The construction of Error Notebooks consists of two steps: (1) collecting historical erroneous CoTs and their incorrect answers, and (2) connecting these CoTs through reflective corrections until the correct solutions are obtained. As a result, the Error Notebooks serve as a repository of tasks along with their corrected CoTs and final answers. RAG is then employed to retrieve specification-relevant records from the Error Notebooks and incorporate them into the inference process. Another major contribution of our work is a human-in-the-loop CAD dataset, which is used to evaluate our method. In addition, the engineering value of our novel framework lies in its ability to effectively handle 3D models with lengthy, non-natural language metadata. Experiments with proprietary models, including GPT-4o and the Gemini series, show substantial gains, with GPT-4o (Omni) achieving up to a 23.4% absolute accuracy improvement on the human preference dataset. Moreover, ablation studies confirm that CoT reasoning provides benefits especially in challenging cases with higher part counts (>10).
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 10:39:37 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 02:22:16 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Liu",
"Yunqing",
""
],
[
"Zhang",
"Nan",
""
],
[
"Tan",
"Zhiming",
""
]
] |
2509.01563
|
Yi-Fan Zhang
|
Biao Yang, Bin Wen, Boyang Ding, Changyi Liu, Chenglong Chu, Chengru Song, Chongling Rao, Chuan Yi, Da Li, Dunju Zang, Fan Yang, Guorui Zhou, Guowang Zhang, Han Shen, Hao Peng, Haojie Ding, Hao Wang, Haonan Fan, Hengrui Ju, Jiaming Huang, Jiangxia Cao, Jiankang Chen, Jingyun Hua, Kaibing Chen, Kaiyu Jiang, Kaiyu Tang, Kun Gai, Muhao Wei, Qiang Wang, Ruitao Wang, Sen Na, Shengnan Zhang, Siyang Mao, Sui Huang, Tianke Zhang, Tingting Gao, Wei Chen, Wei Yuan, Xiangyu Wu, Xiao Hu, Xingyu Lu, Yi-Fan Zhang, Yiping Yang, Yulong Chen, Zeyi Lu, Zhenhua Wu, Zhixin Ling, Zhuoran Yang, Ziming Li, Di Xu, Haixuan Gao, Hang Li, Jing Wang, Lejian Ren, Qigen Hu, Qianqian Wang, Shiyao Wang, Xinchen Luo, Yan Li, Yuhang Hu, Zixing Zhang
|
Kwai Keye-VL 1.5 Technical Report
|
Github page: https://github.com/Kwai-Keye/Keye
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
In recent years, the development of Large Language Models (LLMs) has significantly advanced, extending their capabilities to multimodal tasks through Multimodal Large Language Models (MLLMs). However, video understanding remains a challenging area due to the dynamic and information-dense nature of videos. Existing models struggle with the trade-off between spatial resolution and temporal coverage when processing video content. We present Keye-VL-1.5, which addresses fundamental challenges in video comprehension through three key innovations. First, we introduce a novel Slow-Fast video encoding strategy that dynamically allocates computational resources based on inter-frame similarity, processing key frames with significant visual changes at higher resolution (Slow pathway) while handling relatively static frames with increased temporal coverage at lower resolution (Fast pathway). Second, we implement a progressive four-stage pre-training methodology that systematically extends the model's context length from 8K to 128K tokens, enabling processing of longer videos and more complex visual content. Third, we develop a comprehensive post-training pipeline focusing on reasoning enhancement and human preference alignment, incorporating a 5-step chain-of-thought data construction process, iterative GSPO-based reinforcement learning with progressive prompt hinting for difficult cases, and alignment training. Through extensive evaluation on public benchmarks and rigorous internal human assessment, Keye-VL-1.5 demonstrates significant improvements over existing models, particularly excelling in video understanding tasks while maintaining competitive performance on general multimodal benchmarks.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 15:46:58 GMT"
},
{
"version": "v2",
"created": "Thu, 04 Sep 2025 03:24:09 GMT"
},
{
"version": "v3",
"created": "Sun, 07 Sep 2025 13:40:44 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yang",
"Biao",
""
],
[
"Wen",
"Bin",
""
],
[
"Ding",
"Boyang",
""
],
[
"Liu",
"Changyi",
""
],
[
"Chu",
"Chenglong",
""
],
[
"Song",
"Chengru",
""
],
[
"Rao",
"Chongling",
""
],
[
"Yi",
"Chuan",
""
],
[
"Li",
"Da",
""
],
[
"Zang",
"Dunju",
""
],
[
"Yang",
"Fan",
""
],
[
"Zhou",
"Guorui",
""
],
[
"Zhang",
"Guowang",
""
],
[
"Shen",
"Han",
""
],
[
"Peng",
"Hao",
""
],
[
"Ding",
"Haojie",
""
],
[
"Wang",
"Hao",
""
],
[
"Fan",
"Haonan",
""
],
[
"Ju",
"Hengrui",
""
],
[
"Huang",
"Jiaming",
""
],
[
"Cao",
"Jiangxia",
""
],
[
"Chen",
"Jiankang",
""
],
[
"Hua",
"Jingyun",
""
],
[
"Chen",
"Kaibing",
""
],
[
"Jiang",
"Kaiyu",
""
],
[
"Tang",
"Kaiyu",
""
],
[
"Gai",
"Kun",
""
],
[
"Wei",
"Muhao",
""
],
[
"Wang",
"Qiang",
""
],
[
"Wang",
"Ruitao",
""
],
[
"Na",
"Sen",
""
],
[
"Zhang",
"Shengnan",
""
],
[
"Mao",
"Siyang",
""
],
[
"Huang",
"Sui",
""
],
[
"Zhang",
"Tianke",
""
],
[
"Gao",
"Tingting",
""
],
[
"Chen",
"Wei",
""
],
[
"Yuan",
"Wei",
""
],
[
"Wu",
"Xiangyu",
""
],
[
"Hu",
"Xiao",
""
],
[
"Lu",
"Xingyu",
""
],
[
"Zhang",
"Yi-Fan",
""
],
[
"Yang",
"Yiping",
""
],
[
"Chen",
"Yulong",
""
],
[
"Lu",
"Zeyi",
""
],
[
"Wu",
"Zhenhua",
""
],
[
"Ling",
"Zhixin",
""
],
[
"Yang",
"Zhuoran",
""
],
[
"Li",
"Ziming",
""
],
[
"Xu",
"Di",
""
],
[
"Gao",
"Haixuan",
""
],
[
"Li",
"Hang",
""
],
[
"Wang",
"Jing",
""
],
[
"Ren",
"Lejian",
""
],
[
"Hu",
"Qigen",
""
],
[
"Wang",
"Qianqian",
""
],
[
"Wang",
"Shiyao",
""
],
[
"Luo",
"Xinchen",
""
],
[
"Li",
"Yan",
""
],
[
"Hu",
"Yuhang",
""
],
[
"Zhang",
"Zixing",
""
]
] |
2509.01677
|
Rafael Boto
|
Rafael Boto, Tiago P. Rebelo, Jorge C. Rom\~ao, Jo\~ao P. Silva
|
Machine Learning in the 2HDM2S model for Dark Matter
|
37 pages, 10 figures
| null | null | null |
hep-ph
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We introduce a two real scalar singlet extension of the two Higgs doublet model. We study the vacuum structure, the bounded from below conditions, the restrictions from the oblique parameters S,T and U, as well as the unitarity constraints. We submit the model to collider and Dark Matter experimental constraints and explore its allowed parameter space. We compare randomly populated simulations, simulations starting near the alignment limit, and a Machine Learning based exploration. Using Evolutionary Strategies, we efficiently search for regions with two Dark Matter candidates.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 18:00:32 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 13:31:55 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Boto",
"Rafael",
""
],
[
"Rebelo",
"Tiago P.",
""
],
[
"Romão",
"Jorge C.",
""
],
[
"Silva",
"João P.",
""
]
] |
2509.01694
|
Quang Minh Nguyen
|
Quang Minh Nguyen and Eytan Modiano
|
A QoS Framework for Service Provision in Multi-Infrastructure-Sharing Networks
|
Accepted to ACM MobiHoc '25
| null |
10.1145/3704413.3764414
| null |
cs.NI cs.SY eess.SY
|
http://creativecommons.org/licenses/by/4.0/
|
We propose a framework for resource provisioning with QoS guarantees in shared infrastructure networks. Our novel framework provides tunable probabilistic service guarantees for throughput and delay. Key to our approach is a Modified Dirft-plus-Penalty (MDP) policy that ensures long-term stability while capturing short-term probabilistic service guarantees using linearized upper-confidence bounds. We characterize the feasible region of service guarantees and show that our MDP procedure achieves mean rate stability and an optimality gap that vanishes with the frame size over which service guarantees are provided. Finally, empirical simulations validate our theory and demonstrate the favorable performance of our algorithm in handling QoS in multi-infrastructure networks.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 18:08:33 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 21:19:41 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Nguyen",
"Quang Minh",
""
],
[
"Modiano",
"Eytan",
""
]
] |
2509.01816
|
Christian Wuthrich
|
Marusia Rebolledo and Christian Wuthrich
|
Computing with necklaces on elliptic curves
| null | null | null | null |
math.NT
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
We present computational algorithms to work with points on the modular curve associated to the normaliser of a non-split Cartan group of prime level $p$. Rather than working with explicit equations, we represent these points using the moduli interpretation of necklaces in the $p$-torsion of elliptic curves. We use our methods to investigate for which primes $\ell\neq p$ two rational points with complex multiplication can have equal reduction modulo $\ell$.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 22:47:24 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 16:43:51 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Rebolledo",
"Marusia",
""
],
[
"Wuthrich",
"Christian",
""
]
] |
2509.01837
|
Yotam Erel
|
Yotam Erel, Rishabh Dabral, Vladislav Golyanik, Amit H. Bermano, Christian Theobalt
|
PractiLight: Practical Light Control Using Foundational Diffusion Models
|
Project page: https://yoterel.github.io/PractiLight-project-page/
| null | null | null |
cs.CV
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Light control in generated images is a difficult task, posing specific challenges, spanning over the entire image and frequency spectrum. Most approaches tackle this problem by training on extensive yet domain-specific datasets, limiting the inherent generalization and applicability of the foundational backbones used. Instead, PractiLight is a practical approach, effectively leveraging foundational understanding of recent generative models for the task. Our key insight is that lighting relationships in an image are similar in nature to token interaction in self-attention layers, and hence are best represented there. Based on this and other analyses regarding the importance of early diffusion iterations, PractiLight trains a lightweight LoRA regressor to produce the direct irradiance map for a given image, using a small set of training images. We then employ this regressor to incorporate the desired lighting into the generation process of another image using Classifier Guidance. This careful design generalizes well to diverse conditions and image domains. We demonstrate state-of-the-art performance in terms of quality and control with proven parameter and data efficiency compared to leading works over a wide variety of scenes types. We hope this work affirms that image lighting can feasibly be controlled by tapping into foundational knowledge, enabling practical and general relighting.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 23:38:40 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 03:12:13 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Erel",
"Yotam",
""
],
[
"Dabral",
"Rishabh",
""
],
[
"Golyanik",
"Vladislav",
""
],
[
"Bermano",
"Amit H.",
""
],
[
"Theobalt",
"Christian",
""
]
] |
2509.01840
|
Weicao Deng
|
Weicao Deng, Sangwoo Park, Min Li, and Osvaldo Simeone
|
Optimizing In-Context Learning for Efficient Full Conformal Prediction
|
5 pages, 3 figures
| null | null | null |
cs.LG
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Reliable uncertainty quantification is critical for trustworthy AI. Conformal Prediction (CP) provides prediction sets with distribution-free coverage guarantees, but its two main variants face complementary limitations. Split CP (SCP) suffers from data inefficiency due to dataset partitioning, while full CP (FCP) improves data efficiency at the cost of prohibitive retraining complexity. Recent approaches based on meta-learning or in-context learning (ICL) partially mitigate these drawbacks. However, they rely on training procedures not specifically tailored to CP, which may yield large prediction sets. We introduce an efficient FCP framework, termed enhanced ICL-based FCP (E-ICL+FCP), which employs a permutation-invariant Transformer-based ICL model trained with a CP-aware loss. By simulating the multiple retrained models required by FCP without actual retraining, E-ICL+FCP preserves coverage while markedly reducing both inefficiency and computational overhead. Experiments on synthetic and real tasks demonstrate that E-ICL+FCP attains superior efficiency-coverage trade-offs compared to existing SCP and FCP baselines.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 23:49:27 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 15:24:37 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Deng",
"Weicao",
""
],
[
"Park",
"Sangwoo",
""
],
[
"Li",
"Min",
""
],
[
"Simeone",
"Osvaldo",
""
]
] |
2509.01881
|
Shurui Lin
|
Shurui Lin, Yin Li, Shy Genel, Francisco Villaescusa-Navarro, Biwei Dai, Wentao Luo, and Yang Wang
|
One latent to fit them all: a unified representation of baryonic feedback on matter distribution
|
10 pages and 5 figures in the main text; 9 pages, 5 figures, and 3 tables in the appendix
| null | null | null |
astro-ph.CO astro-ph.IM
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Accurate and parsimonious quantification of baryonic feedback on matter distribution is of crucial importance for understanding both cosmology and galaxy formation from observational data. This is, however, challenging given the large discrepancy among different models of galaxy formation simulations, and their distinct subgrid physics parameterizations. Using 5,072 simulations from 4 different models covering broad ranges in their parameter spaces, we find a unified 2D latent representation. Compared to the simulations and other phenomenological models, our representation is independent of both time and cosmology, much lower-dimensional, and disentangled in its impacts on the matter power spectra. The common latent space facilitates the comparison of parameter spaces of different models and is readily interpretable by correlation with each. The two latent dimensions provide a complementary representation of baryonic effects, linking black hole and supernova feedback to distinct and interpretable impacts on the matter power spectrum. Our approach enables developing robust and economical analytic models for optimal gain of physical information from data, and is generalizable to other fields with significant modeling uncertainty.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 02:12:06 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 23:05:08 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Lin",
"Shurui",
""
],
[
"Li",
"Yin",
""
],
[
"Genel",
"Shy",
""
],
[
"Villaescusa-Navarro",
"Francisco",
""
],
[
"Dai",
"Biwei",
""
],
[
"Luo",
"Wentao",
""
],
[
"Wang",
"Yang",
""
]
] |
2509.01901
|
Alexander Clow A.Clow
|
Saieed Akbari and Jonny Aloni and Arash Beikmohammadi and Alexander Clow
|
Tight Bounds for Cycle-Edge Decompositions and Covers
|
12 pages, 1 figure
| null | null | null |
math.CO
|
http://creativecommons.org/licenses/by/4.0/
|
An old conjecture of Erd{\H{o}}s and Gallai states that every $n$ vertex graph can be decomposed, that is $E(G)$ can be partitioned, into $O(n)$ cycles and edges. The covering version of this conjecture was proven by Pyber in 1985, where it was shown that all graphs can be covered by $n-1$ cycles and edges. The best upper bound on the number of cycles and edges required to decompose any graph is $O(n\log^*(n))$, which was recently shown by Buci{\'c} and Montgomery in 2023. Here $\log^*(n)$ denotes the iterated logarithm function. Meanwhile, a construction of Erd\H{o}s demonstrate that there exists graphs which require $(\frac{3}{2}-o(1))n$ cycles and edges to be decomposed. We prove all graphs with maximum degree at most $4$ can be decomposed into $n-1$ or fewer cycles and edges. We also show that every $n$ vertex claw-free graph can be decomposed into $n-1$ or fewer $2$-regular subgraphs and edges. Finally, we prove that every graph $G$ containing a cycle can be covered by $n-2$ or fewer cycles and edges. This improves Pyber's covering theorem by proving that $n-1$ cycles and edges are required only for trees.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 02:43:29 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 17:15:33 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Akbari",
"Saieed",
""
],
[
"Aloni",
"Jonny",
""
],
[
"Beikmohammadi",
"Arash",
""
],
[
"Clow",
"Alexander",
""
]
] |
2509.01909
|
Ranjie Duan
|
Ranjie Duan, Jiexi Liu, Xiaojun Jia, Shiji Zhao, Ruoxi Cheng, Fengxiang Wang, Cheng Wei, Yong Xie, Chang Liu, Defeng Li, Yinpeng Dong, Yichi Zhang, Yuefeng Chen, Chongwen Wang, Xingjun Ma, Xingxing Wei, Yang Liu, Hang Su, Jun Zhu, Xinfeng Li, Yitong Sun, Jie Zhang, Jinzhao Hu, Sha Xu, Yitong Yang, Jialing Tao, Hui Xue
|
Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Models
|
Technical Report Code & Model weights available: https://github.com/Alibaba-AAIG/Oyster
| null | null | null |
cs.AI cs.CL cs.CY cs.HC cs.SC
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
Large language models (LLMs) typically deploy safety mechanisms to prevent harmful content generation. Most current approaches focus narrowly on risks posed by malicious actors, often framing risks as adversarial events and relying on defensive refusals. However, in real-world settings, risks also come from non-malicious users seeking help while under psychological distress (e.g., self-harm intentions). In such cases, the model's response can strongly influence the user's next actions. Simple refusals may lead them to repeat, escalate, or move to unsafe platforms, creating worse outcomes. We introduce Constructive Safety Alignment (CSA), a human-centric paradigm that protects against malicious misuse while actively guiding vulnerable users toward safe and helpful results. Implemented in Oyster-I (Oy1), CSA combines game-theoretic anticipation of user reactions, fine-grained risk boundary discovery, and interpretable reasoning control, turning safety into a trust-building process. Oy1 achieves state-of-the-art safety among open models while retaining high general capabilities. On our Constructive Benchmark, it shows strong constructive engagement, close to GPT-5, and unmatched robustness on the Strata-Sword jailbreak dataset, nearing GPT-o1 levels. By shifting from refusal-first to guidance-first safety, CSA redefines the model-user relationship, aiming for systems that are not just safe, but meaningfully helpful. We release Oy1, code, and the benchmark to support responsible, user-centered AI.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 03:04:27 GMT"
},
{
"version": "v2",
"created": "Thu, 04 Sep 2025 11:54:06 GMT"
},
{
"version": "v3",
"created": "Mon, 08 Sep 2025 15:18:35 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Duan",
"Ranjie",
""
],
[
"Liu",
"Jiexi",
""
],
[
"Jia",
"Xiaojun",
""
],
[
"Zhao",
"Shiji",
""
],
[
"Cheng",
"Ruoxi",
""
],
[
"Wang",
"Fengxiang",
""
],
[
"Wei",
"Cheng",
""
],
[
"Xie",
"Yong",
""
],
[
"Liu",
"Chang",
""
],
[
"Li",
"Defeng",
""
],
[
"Dong",
"Yinpeng",
""
],
[
"Zhang",
"Yichi",
""
],
[
"Chen",
"Yuefeng",
""
],
[
"Wang",
"Chongwen",
""
],
[
"Ma",
"Xingjun",
""
],
[
"Wei",
"Xingxing",
""
],
[
"Liu",
"Yang",
""
],
[
"Su",
"Hang",
""
],
[
"Zhu",
"Jun",
""
],
[
"Li",
"Xinfeng",
""
],
[
"Sun",
"Yitong",
""
],
[
"Zhang",
"Jie",
""
],
[
"Hu",
"Jinzhao",
""
],
[
"Xu",
"Sha",
""
],
[
"Yang",
"Yitong",
""
],
[
"Tao",
"Jialing",
""
],
[
"Xue",
"Hui",
""
]
] |
2509.01932
|
Victor Vasilevsky S.
|
N. K. Kalzhigitov, S. Amangeldinova, V. O. Kurmangaliyeva, V. S. Vasilevsky
|
Discrete and continuous spectrum of lightest hypernuclei
|
14 pages, 16 figures, typos corrected, text transformed to the style of Chin. J. Phys
| null | null | null |
nucl-th
|
http://creativecommons.org/licenses/by/4.0/
|
We analyze the peculiarities of the interaction of the lambda hyperon with s-shell nuclei. The spectra of bound and scattering states are studied in the hypernuclei $_{\Lambda}^{2}$H, $_{\Lambda}^{3}$H, $_{\Lambda}^{4}$H and $_{\Lambda}^{5}$He, which is considered as two-cluster configurations $p+\Lambda$, $d$+$\Lambda$, $^{3}$H+$\Lambda$, $^{4}$He+$\Lambda$, respectively. The explicit form of the folding potentials of such an interaction is presented in coordinate and oscillator representations, which help us to understand the structure of hypernuclei of interest. We compare energies of bound states, phase shifts of the elastic lambda hyperon, and neutron scattering from s-shell nuclei.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 04:03:33 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 05:00:17 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Kalzhigitov",
"N. K.",
""
],
[
"Amangeldinova",
"S.",
""
],
[
"Kurmangaliyeva",
"V. O.",
""
],
[
"Vasilevsky",
"V. S.",
""
]
] |
2509.01979
|
Huijun Yang
|
Huijun Yang
|
A Bilinear Form for Spin$^c$ Manifolds
| null | null | null | null |
math.AT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Let $M$ be a closed oriented spin$^{c}$ manifold of dimension $(8n {+} 2)$ with fundamental class $[M]$, and let $\rho_{2} \colon H^{4n}(M; \mathbb{Z}) \rightarrow H^{4n}(M; \mathbb{Z}/2)$ denote the $\bmod ~ 2$ reduction homomorphism. For any torsion class $t \in H^{4n}(M;\mathbb{Z})$, we establish the identity \[ \langle \rho_2(t) \cdot Sq^2 \rho_2 (t), [M] \rangle = \langle \rho_2 (t) \cdot Sq^2 v_{4n}(M), [M]\rangle, \] where $Sq^2$ is the Steenrod square, $v_{4n}(M)$ is the $4n$-th Wu class of $M$, $ x\cdot y$ denotes the cup product of $x$ and $y$, and $\langle \cdot ~, ~\cdot \rangle$ denotes the Kronecker product. This result generalizes the work of Landweber and Stong from spin to spin$^c$ manifolds.
As an application, let $\beta^{\mathbb{Z}/2} \colon H^{4n+2}(M; \mathbb{Z}/2) \to H^{4n+3}(M; \mathbb{Z})$ be the Bockstein homomorphism associated to the short exact sequence of coefficients $\mathbb{Z} \xrightarrow{\times 2} \mathbb{Z} \to \mathbb{Z}/2$. We deduce that $\beta^{\mathbb{Z}/2}(Sq^2 v_{4n}(M)) = 0$, and consequently, $Sq^3 v_{4n}(M) = 0$, for any closed oriented spin$^{c}$ manifold $M$ with $\dim M \le 8n{+}1$.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 05:45:36 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 00:20:17 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yang",
"Huijun",
""
]
] |
2509.02100
|
Sharjeel Tahir
|
Sharjeel Tahir, Judith Johnson, Jumana Abu-Khalaf and Syed Afaq Ali Shah
|
E-THER: A Multimodal Dataset for Empathic AI -- Towards Emotional Mismatch Awareness
|
15 pages, 4 figures. Preprint
| null | null | null |
cs.HC cs.CL
|
http://creativecommons.org/licenses/by/4.0/
|
A prevalent shortfall among current empathic AI systems is their inability to recognize when verbal expressions may not fully reflect underlying emotional states. This is because the existing datasets, used for the training of these systems, focus on surface-level emotion recognition without addressing the complex verbal-visual incongruence (mismatch) patterns useful for empathic understanding. In this paper, we present E-THER, the first Person-Centered Therapy-grounded multimodal dataset with multidimensional annotations for verbal-visual incongruence detection, enabling training of AI systems that develop genuine rather than performative empathic capabilities. The annotations included in the dataset are drawn from humanistic approach, i.e., identifying verbal-visual emotional misalignment in client-counsellor interactions - forming a framework for training and evaluating AI on empathy tasks. Additional engagement scores provide behavioral annotations for research applications. Notable gains in empathic and therapeutic conversational qualities are observed in state-of-the-art vision-language models (VLMs), such as IDEFICS and VideoLLAVA, using evaluation metrics grounded in empathic and therapeutic principles. Empirical findings indicate that our incongruence-trained models outperform general-purpose models in critical traits, such as sustaining therapeutic engagement, minimizing artificial or exaggerated linguistic patterns, and maintaining fidelity to PCT theoretical framework.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 08:58:32 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 08:37:27 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Tahir",
"Sharjeel",
""
],
[
"Johnson",
"Judith",
""
],
[
"Abu-Khalaf",
"Jumana",
""
],
[
"Shah",
"Syed Afaq Ali",
""
]
] |
2509.02130
|
Kim Hammar
|
Kim Hammar and Rolf Stadler
|
Online Identification of IT Systems through Active Causal Learning
| null | null | null | null |
cs.LG cs.DC cs.NI
|
http://creativecommons.org/licenses/by-sa/4.0/
|
Identifying a causal model of an IT system is fundamental to many branches of systems engineering and operation. Such a model can be used to predict the effects of control actions, optimize operations, diagnose failures, detect intrusions, etc., which is central to achieving the longstanding goal of automating network and system management tasks. Traditionally, causal models have been designed and maintained by domain experts. This, however, proves increasingly challenging with the growing complexity and dynamism of modern IT systems. In this paper, we present the first principled method for online, data-driven identification of an IT system in the form of a causal model. The method, which we call active causal learning, estimates causal functions that capture the dependencies among system variables in an iterative fashion using Gaussian process regression based on system measurements, which are collected through a rollout-based intervention policy. We prove that this method is optimal in the Bayesian sense and that it produces effective interventions. Experimental validation on a testbed shows that our method enables accurate identification of a causal system model while inducing low interference with system operations.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 09:25:36 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 02:18:18 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Hammar",
"Kim",
""
],
[
"Stadler",
"Rolf",
""
]
] |
2509.02215
|
HyeonSeop Oh
|
Xushan Huang, Hobin Lee, HyeonSeop Oh
|
Stability of viscous shock for the Navier-Stokes-Fourier system: outflow and impermeable wall problems
| null | null | null | null |
math.AP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We investigate the time-asymptotic stability of solutions to the one-dimensional Navier-Stokes-Fourier system in the half space, focusing on the outflow and impermeable wall problems. When the prescribed boundary and far-field conditions form an outgoing viscous shock, we prove that the solution converges to the viscous shock profile, up to a dynamical shift, provided that the initial perturbation and the shock amplitude are sufficiently small. In order to obtain our results, we employ the method of a-contraction with shifts. Although the impermeable wall problem is technically simpler to analyze in Lagrangian mass coordinates, the outflow problem leads to a free boundary in that framework. Therefore, we use Eulerian coordinates to provide a unified approach to both problems. This is the first result on the time-asymptotic stability of viscous shocks for initial-boundary value problems of the Navier-Stokes-Fourier system for the outflow and impermeable wall cases.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 11:34:01 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 06:15:52 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Huang",
"Xushan",
""
],
[
"Lee",
"Hobin",
""
],
[
"Oh",
"HyeonSeop",
""
]
] |
2509.02268
|
Xiehang Yu
|
Xie-Hang Yu, Hongchao Li, J. Ignacio Cirac, and Rahul Trivedi
|
Optimizing digital quantum simulation of open quantum lattice models
|
61 pages, 6 figures
| null | null | null |
quant-ph cond-mat.stat-mech
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Many-body systems arising in condensed matter physics and quantum optics inevitably couple to the environment and need to be modelled as open quantum systems. While near-optimal algorithms have been developed for simulating many-body quantum dynamics, algorithms for their open system counterparts remain less well investigated. We address the problem of simulating geometrically local many-body open quantum systems interacting with a stationary Gaussian environment. Under a smoothness assumption on the system-environment interaction, we develop near-optimal algorithms that, for a model with $N$ spins and evolution time $t$, attain a simulation error $\delta$ in the system-state with $\mathcal{O}(Nt(Nt/\delta)^{o(1)})$ gates, $\mathcal{O}(t(Nt/\delta)^{o(1)})$ parallelized circuit depth and $\tilde{\mathcal{O}}(N(Nt/\delta)^{o(1)})$ ancillas. We additionally show that, if only simulating local observables is of interest, then the circuit depth of the digital algorithm can be chosen to be independent of the system size $N$. This provides theoretical evidence for the utility of these algorithms for simulating physically relevant models, where typically local observables are of interest, on pre-fault tolerant devices. Finally, for the limiting case of Markovian dynamics with commuting jump operators, we propose two algorithms based on sampling a Wiener process and on a locally dilated Hamiltonian construction, respectively. These algorithms reduce the asymptotic gate complexity on $N$ compared to currently available algorithms in terms of the required number of geometrically local gates.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 12:44:57 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 10:21:38 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yu",
"Xie-Hang",
""
],
[
"Li",
"Hongchao",
""
],
[
"Cirac",
"J. Ignacio",
""
],
[
"Trivedi",
"Rahul",
""
]
] |
2509.02333
|
Shihui Yang
|
Shihui Yang, Chengfeng Dou, Peidong Guo, Kai Lu, Qiang Ju, Fei Deng, Rihui Xin
|
DCPO: Dynamic Clipping Policy Optimization
| null | null | null | null |
cs.CL cs.AI cs.LG
|
http://creativecommons.org/licenses/by/4.0/
|
Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as a promising framework for enhancing the reasoning capabilities of large language models. However, existing approaches such as GRPO often suffer from zero gradients. This problem arises primarily due to fixed clipping bounds for token-level probability ratios and the standardization of identical rewards, which can lead to ineffective gradient updates and underutilization of generated responses. In this work, we propose Dynamic Clipping Policy Optimization(DCPO), which introduces a dynamic clipping strategy that adaptively adjusts clipping bounds based on token-specific prior probabilities to enhance token-level exploration, and a smooth advantage standardization technique that standardizes rewards across cumulative training steps to improve the response-level effective utilization of generated responses. DCPO achieved state-of-the-art performance on four benchmarks based on four different models. In particular, DCPO achieved an Avg@1 of 46.7 under greedy decoding and an Avg@32 of 38.8 under 32 times sampling on the AIME24 benchmark, surpassing DAPO (36.7/31.6), GRPO (36.7/32.1) and GSPO (40.0/34.9) on the Qwen2.5-Math-7B model. On the AIME25 benchmark based on Qwen2.5-14B, DCPO achieves a performance of (23.3/19.0), surpassing GRPO (13.3/10.5), DAPO (20.0/15.3) and GSPO (16.7/9.9). Furthermore, DCPO achieved an average 28% improvement in the nonzero advantage over GRPO in four models, doubled the training efficiency over DAPO, and significantly reduced the token clipping ratio by an order of magnitude compared to both GRPO and DAPO, while achieving superior performance. These results highlight DCPO's effectiveness in leveraging generated data more efficiently for reinforcement learning in large language models.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 14:01:07 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 14:50:44 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yang",
"Shihui",
""
],
[
"Dou",
"Chengfeng",
""
],
[
"Guo",
"Peidong",
""
],
[
"Lu",
"Kai",
""
],
[
"Ju",
"Qiang",
""
],
[
"Deng",
"Fei",
""
],
[
"Xin",
"Rihui",
""
]
] |
2509.02335
|
Vaibhav Raj Singh Parmar PhD Candidate
|
Vaibhav Raj Singh Parmar and Ranjini Bandyopadhyay
|
Permeability heterogeneity and bulk linear elasticity of displaced clay suspensions determine interfacial pattern morphologies in Hele-Shaw experiments
|
21 pages including supplementary information, 7 figures, Supplementary videos included with the source files
| null | null | null |
cond-mat.soft
|
http://creativecommons.org/licenses/by/4.0/
|
When a less viscous Newtonian fluid displaces an aging aqueous clay suspension in a confined space, a rich array of interfacial patterns emerges due to a predominantly viscous instability. In the present work, we controlled the mechanical properties of clay suspensions by incorporating additives and studied the interfacial instabilities that resulted when these suspensions were radially displaced by water in a Hele-Shaw cell. When the elasticity of clay was low, the interfacial dynamics exhibited features of nonlinear viscous fingering in heterogeneous media. By tuning the nature and content of additives that delay clay aging, we uncovered two novel propagation mechanisms: pattern growth \textit{via} skewering and zig-zag finger propagation. These patterns have hitherto never been observed in experiments with colloidal systems. For moderate clay elasticities, we demonstrate here that shear-thinning-induced flow anisotropy leads to the formation of dendrites with dominant side branches. As clay elasticity increases due to the incorporation of salts, the energy required to create fractures becomes smaller than that for system-wide yielding. This scenario is characterized by the emergence of viscoelastic fractures. Our work demonstrates that incorporating additives is an effective strategy to manipulate the onset and growth of interfacial instabilities during the confined displacement of clay by miscible Newtonian fluids.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 14:03:43 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 08:39:00 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Parmar",
"Vaibhav Raj Singh",
""
],
[
"Bandyopadhyay",
"Ranjini",
""
]
] |
2509.02410
|
Iosif Pinelis
|
Iosif Pinelis
|
Comment on "Worst-case Nonparametric Bounds for the Student T-statistic", arXiv:2508.13226
|
1 page; comment on arXiv:2508.13226. Version 2: two typos corrected, presentation slightly improved. Version 3: added a comment on Version 2 of arXiv:2508.13226
| null | null | null |
math.ST math.PR stat.TH
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Concerning Version 1 of ``Worst-case Nonparametric Bounds for the Student T-statistic'', arXiv:2508.13226: The main result there is incorrect. Concerning Version 2 of arXiv:2508.13226: At least the proof of the main result there is incorrect.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 15:19:28 GMT"
},
{
"version": "v2",
"created": "Wed, 03 Sep 2025 15:52:19 GMT"
},
{
"version": "v3",
"created": "Fri, 05 Sep 2025 19:48:21 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Pinelis",
"Iosif",
""
]
] |
2509.02499
|
Junxi Wu
|
Junxi Wu and Jinpeng Wang and Zheng Liu and Bin Chen and Dongjian Hu and Hao Wu and Shu-Tao Xia
|
MoSEs: Uncertainty-Aware AI-Generated Text Detection via Mixture of Stylistics Experts with Conditional Thresholds
|
EMNLP 2025
| null | null | null |
cs.CL cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
The rapid advancement of large language models has intensified public concerns about the potential misuse. Therefore, it is important to build trustworthy AI-generated text detection systems. Existing methods neglect stylistic modeling and mostly rely on static thresholds, which greatly limits the detection performance. In this paper, we propose the Mixture of Stylistic Experts (MoSEs) framework that enables stylistics-aware uncertainty quantification through conditional threshold estimation. MoSEs contain three core components, namely, the Stylistics Reference Repository (SRR), the Stylistics-Aware Router (SAR), and the Conditional Threshold Estimator (CTE). For input text, SRR can activate the appropriate reference data in SRR and provide them to CTE. Subsequently, CTE jointly models the linguistic statistical properties and semantic features to dynamically determine the optimal threshold. With a discrimination score, MoSEs yields prediction labels with the corresponding confidence level. Our framework achieves an average improvement 11.34% in detection performance compared to baselines. More inspiringly, MoSEs shows a more evident improvement 39.15% in the low-resource case. Our code is available at https://github.com/creator-xi/MoSEs.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 16:51:43 GMT"
},
{
"version": "v2",
"created": "Wed, 03 Sep 2025 02:52:22 GMT"
},
{
"version": "v3",
"created": "Mon, 08 Sep 2025 02:08:49 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Wu",
"Junxi",
""
],
[
"Wang",
"Jinpeng",
""
],
[
"Liu",
"Zheng",
""
],
[
"Chen",
"Bin",
""
],
[
"Hu",
"Dongjian",
""
],
[
"Wu",
"Hao",
""
],
[
"Xia",
"Shu-Tao",
""
]
] |
2509.02539
|
Ivonne Zavala
|
Daniele Licciardello, Saba Rahimy, Ivonne Zavala
|
Extending the Dynamical Systems Toolkit: Coupled Fields in Multiscalar Dark Energy
|
36 pages, 1 figure
| null | null | null |
hep-th gr-qc
|
http://creativecommons.org/licenses/by/4.0/
|
We study the dynamics of a two-field scalar model consisting of an axion-saxion pair with both kinetic and potential couplings, as motivated by string theory compactifications. We extend the dynamical systems (DS) toolkit by introducing a new set of variables that not only close the system and enable a systematic stability analysis, but also disentangle the role of the kinetic coupling. Within this framework we derive a compact, general expression for the non-geodesicity (turning-rate) parameter evaluated at fixed points, valid for arbitrary couplings. This provides a transparent way of diagnosing non-geodesic dynamics, with direct applications to both dark energy and multifield inflation. We first consider exponential coupling functions to establish analytic control and facilitate comparison with previous literature. In this case, we uncover a pair of genuinely non-geodesic fixed points, which act as attractors within a submanifold of the full system. In contrast, when the axion shift symmetry remains unbroken, our analysis shows that the apparent non-geodesic fixed point reported previously does not persist once the full dynamics are taken into account. Finally, we illustrate how our approach naturally extends to more realistic string-inspired models, such as power-law axion potentials combined with exponential saxion couplings, and present an explicit supergravity realisation.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 17:41:58 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Licciardello",
"Daniele",
""
],
[
"Rahimy",
"Saba",
""
],
[
"Zavala",
"Ivonne",
""
]
] |
2509.02625
|
Xiaoyu Yan
|
Xiaoyu Yan
|
The Recursive Panopticon Hypothesis
|
This manuscript is being withdrawn at the author's request due to concerns regarding the improper use of AI tools in its preparation, which does not align with standards of academic transparency. The author takes full responsibility for this oversight
| null | null | null |
physics.soc-ph
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
The cosmic "Great Silence" revealed by the Fermi paradox remains a central puzzle in contemporary science. Existing explanations such as the "Big Filter," "Zoo Hypothesis," and "Dark Forest" theory are trapped in isolated frameworks of "hypothesis list paradigm" that resist falsification. This paper proposes the "Recursive Panopticon Hypothesis" arguing that under the uncertainty of recursive higher-order deterrence, cosmic civilizations will universally adopt "silence" as their optimal survival strategy through rational risk avoidance. To test this hypothesis, we innovatively introduce the interdisciplinary research paradigm of "Computational Cosmic Sociology." By constructing a highly parameterized Agent-Based Modeling (ABM) simulation, we abstract civilizations as rational agents with risk perception, strategy learning, and interactive memory evolving within a simulated cosmic grid. The model's core lies in a utility function based on recursive deterrence theory and a network co-evolution mechanism connecting micro-decisions with macro-social structures. Research findings indicate: "Silence" is an evolutionarily stable strategy; the "Dark Forest" state is merely a special case of system instability under extreme resource scarcity and high-density civilizations; civilizational interactions spontaneously form structured social networks with small-world properties; and a hypothetical "Ultimate Civilization" can effectively maintain order. This study aims to drive paradigm shifts, from listing mutually exclusive hypotheses to a unified, computable theoretical framework, thereby establishing an empirical foundation for cosmological sociology and providing profound insights for SETI strategies.
|
[
{
"version": "v1",
"created": "Mon, 01 Sep 2025 14:35:16 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 11:34:14 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Yan",
"Xiaoyu",
""
]
] |
2509.02653
|
Xiangpeng Li
|
Xiangpeng Li, Junwei Ma, Bo Li, Ali Mostafavi
|
Quantifying the Social Costs of Power Outages and Restoration Disparities Across Four U.S. Hurricanes
| null | null | null | null |
physics.soc-ph cs.LG econ.GN q-fin.EC
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
The multifaceted nature of disaster impact shows that densely populated areas contribute more to aggregate burden, while sparsely populated but heavily affected regions suffer disproportionately at the individual level. This study introduces a framework for quantifying the societal impacts of power outages by translating customer weighted outage exposure into deprivation measures, integrating welfare metrics with three recovery indicators, average outage days per customer, restoration duration, and relative restoration rate, computed from sequential EAGLE I observations and linked to Zip Code Tabulation Area demographics. Applied to four United States hurricanes, Beryl 2024 Texas, Helene 2024 Florida, Milton 2024 Florida, and Ida 2021 Louisiana, this standardized pipeline provides the first cross event, fine scale evaluation of outage impacts and their drivers. Results demonstrate regressive patterns with greater burdens in lower income areas, mechanistic analysis shows deprivation increases with longer restoration durations and decreases with faster restoration rates, explainable modeling identifies restoration duration as the dominant driver, and clustering reveals distinct recovery typologies not captured by conventional reliability metrics. This framework delivers a transferable method for assessing outage impacts and equity, comparative cross event evidence linking restoration dynamics to social outcomes, and actionable spatial analyses that support equity informed restoration planning and resilience investment.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 14:32:08 GMT"
},
{
"version": "v2",
"created": "Fri, 05 Sep 2025 18:29:41 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Li",
"Xiangpeng",
""
],
[
"Ma",
"Junwei",
""
],
[
"Li",
"Bo",
""
],
[
"Mostafavi",
"Ali",
""
]
] |
2509.02722
|
Delong Chen
|
Delong Chen, Theo Moutakanni, Willy Chung, Yejin Bang, Ziwei Ji, Allen Bolourchi, Pascale Fung
|
Planning with Reasoning using Vision Language World Model
| null | null | null | null |
cs.AI
|
http://creativecommons.org/licenses/by/4.0/
|
Effective planning requires strong world models, but high-level world models that can understand and reason about actions with semantic and temporal abstraction remain largely underdeveloped. We introduce the Vision Language World Model (VLWM), a foundation model trained for language-based world modeling on natural videos. Given visual observations, the VLWM first infers the overall goal achievements then predicts a trajectory composed of interleaved actions and world state changes. Those targets are extracted by iterative LLM Self-Refine conditioned on compressed future observations represented by Tree of Captions. The VLWM learns both an action policy and a dynamics model, which respectively facilitates reactive system-1 plan decoding and reflective system-2 planning via cost minimization. The cost evaluates the semantic distance between the hypothetical future states given by VLWM roll-outs and the expected goal state, and is measured by a critic model that we trained in a self-supervised manner. The VLWM achieves state-of-the-art Visual Planning for Assistance (VPA) performance on both benchmark evaluations and our proposed PlannerArena human evaluations, where system-2 improves the Elo score by +27% upon system-1. The VLWM models also outperforms strong VLM baselines on RoboVQA and WorldPrediction benchmark.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 18:18:57 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 21:02:02 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Chen",
"Delong",
""
],
[
"Moutakanni",
"Theo",
""
],
[
"Chung",
"Willy",
""
],
[
"Bang",
"Yejin",
""
],
[
"Ji",
"Ziwei",
""
],
[
"Bolourchi",
"Allen",
""
],
[
"Fung",
"Pascale",
""
]
] |
2509.02769
|
Tom\'as Ahumada
|
Vishwajeet Swain, Tom\'as Ahumada, Sameer K. Patil, Yogesh Wagh, Varun Bhalerao, Ehud Nakar, Mansi Kasliwal, Xander J. Hall, Malte Busmann, Shreya Anand, Viraj Karambelkar, Igor Andreoni, G. C. Anupama, Anuraag Arya, Arvind Balasubramanian, Sudhanshu Barway, Jonathan Carney, Michael Coughlin, Deepak Eappachen, James Freeburn, Daniel Gruen, Tanishk Mohan, Brendan O'Connor, Antonella Palmese, Utkarsh Pathak, D. K. Sahu, Aditya Pawan Saikia, Nikhil Sarin, Gokul Srinivasaragavan, and Hitesh Tanenia
|
GRB 250704B: An Off-axis Short GRB with a Long-Lived Afterglow Plateau
|
submitted
| null | null | null |
astro-ph.HE
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
We present a detailed multi-wavelength afterglow study of the short GRB 250704B, extensively monitored in optical and near-infrared bands. Its afterglow displays an unusually long-duration plateau followed by an achromatic break and a steep decline, deviating from canonical GRB afterglows. While long plateaus are often explained by central engine activity, we find that for GRB 250704B, an energy injection model requires unreasonable parameters. The afterglow is better explained by an off-axis power-law structured jet with a narrow core ($\theta_c \approx 0.7^{\circ}$) viewed at a modest angle ($\theta_v \approx 1.9^{\circ}$). A comparison with GRB 170817A shows that both events are consistent with the off-axis structured jet scenario, where the shape of the light curve is governed primarily by the geometry of the jet and the viewing angle rather than the energetics, microphysical parameters, or external density. Our results underscore the importance of incorporating the jet structure in GRB modeling.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 19:17:13 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 17:55:38 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Swain",
"Vishwajeet",
""
],
[
"Ahumada",
"Tomás",
""
],
[
"Patil",
"Sameer K.",
""
],
[
"Wagh",
"Yogesh",
""
],
[
"Bhalerao",
"Varun",
""
],
[
"Nakar",
"Ehud",
""
],
[
"Kasliwal",
"Mansi",
""
],
[
"Hall",
"Xander J.",
""
],
[
"Busmann",
"Malte",
""
],
[
"Anand",
"Shreya",
""
],
[
"Karambelkar",
"Viraj",
""
],
[
"Andreoni",
"Igor",
""
],
[
"Anupama",
"G. C.",
""
],
[
"Arya",
"Anuraag",
""
],
[
"Balasubramanian",
"Arvind",
""
],
[
"Barway",
"Sudhanshu",
""
],
[
"Carney",
"Jonathan",
""
],
[
"Coughlin",
"Michael",
""
],
[
"Eappachen",
"Deepak",
""
],
[
"Freeburn",
"James",
""
],
[
"Gruen",
"Daniel",
""
],
[
"Mohan",
"Tanishk",
""
],
[
"O'Connor",
"Brendan",
""
],
[
"Palmese",
"Antonella",
""
],
[
"Pathak",
"Utkarsh",
""
],
[
"Sahu",
"D. K.",
""
],
[
"Saikia",
"Aditya Pawan",
""
],
[
"Sarin",
"Nikhil",
""
],
[
"Srinivasaragavan",
"Gokul",
""
],
[
"Tanenia",
"Hitesh",
""
]
] |
2509.02784
|
M\'aria Lakatos
|
M\'aria Lakatos
|
A Composite-Loss Graph Neural Network for the Multivariate Post-Processing of Ensemble Weather Forecasts
|
30 pages, 16 figures, 3 tables
| null | null | null |
stat.AP stat.ML
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Ensemble forecasting systems have advanced meteorology by providing probabilistic estimates of future states. Nonetheless, systematic biases often persist, making statistical post-processing essential. Traditional parametric post-processing techniques and machine learning-based methods can produce calibrated predictive distributions at specific locations and lead times, yet often struggle to capture dependencies across forecast dimensions. To address this, multivariate post-processing methods-such as ensemble copula coupling and the Schaake shuffle-are widely applied in a second step to restore realistic inter-variable or spatio-temporal dependencies. The aim of this study is the multivariate post-processing of ensemble forecasts using a graph neural network (dualGNN) trained with a composite loss function that combines the energy score (ES) and the variogram score (VS). The method is evaluated on two datasets: WRF-based solar irradiance forecasts over northern Chile and ECMWF visibility forecasts for Central Europe. The dualGNN consistently outperforms all empirical copula-based post-processed forecasts and shows significant improvements compared to graph neural networks trained solely on either the continuous ranked probability score or the ES, according to the evaluated multivariate verification metrics. Furthermore, for the WRF forecasts, the rank-order structure of the dualGNN forecasts captures valuable dependency information, enabling a more effective restoration of spatial relationships than either the raw numerical weather prediction ensemble or historical observational rank structures. Notably, incorporating VS into the loss function improved the univariate performance for both target variables compared to training on ES alone. Moreover, for the visibility forecasts, the ES-VS combination even outperformed the strongest calibrated reference in terms of univariate performance.
|
[
{
"version": "v1",
"created": "Tue, 02 Sep 2025 19:37:44 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 14:52:18 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Lakatos",
"Mária",
""
]
] |
2509.02929
|
Jordan Brook
|
Jordan Brook, Hamish McGowan, Matt Mason, Joshua Soderholm and Alain Protat
|
The Australian Identification, Nowcasting and Tracking Algorithm (A.I.N.T.)
|
This document has been uploaded upon request to create a reference for subsequent work that has used it
| null | null | null |
physics.ao-ph
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
This is a reupload of an industry funded research project description from 2019. The project heavily relied upon the TINT is not Titan package (TINT, available at https://github.com/openradar/TINT). This document details the modifications made to TINT and its basic functionality, along with the a description of the resulting database created for industry partners.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 01:45:23 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 06:17:21 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Brook",
"Jordan",
""
],
[
"McGowan",
"Hamish",
""
],
[
"Mason",
"Matt",
""
],
[
"Soderholm",
"Joshua",
""
],
[
"Protat",
"Alain",
""
]
] |
2509.02981
|
Minxin Zhang
|
Minxin Zhang, Yuxuan Liu, Hayden Schaeffer
|
AdaGrad Meets Muon: Adaptive Stepsizes for Orthogonal Updates
| null | null | null | null |
cs.LG math.OC
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The recently proposed Muon optimizer updates weight matrices via orthogonalized momentum and has demonstrated strong empirical success in large language model training. However, it remains unclear how to determine the learning rates for such orthogonalized updates. AdaGrad, by contrast, is a widely used adaptive method that scales stochastic gradients by accumulated past gradients. We propose a new algorithm, AdaGO, which combines a norm-based AdaGrad-type stepsize with an orthogonalized update direction, bringing together the benefits of both approaches. Unlike other adaptive variants of Muon, AdaGO preserves the orthogonality of the update direction, which can be interpreted as a spectral descent direction, while adapting the stepsizes to the optimization landscape by scaling the direction with accumulated past gradient norms. The implementation of AdaGO requires only minimal modification to Muon, with a single additional scalar variable, the accumulated squared gradient norms, to be computed, making it computationally and memory efficient. Optimal theoretical convergence rates are established for nonconvex functions in both stochastic and deterministic settings under standard smoothness and unbiased bounded-variance noise assumptions. Empirical results on CIFAR-10 classification and function regression demonstrate that AdaGO outperforms Muon and Adam.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 03:42:22 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 21:55:53 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Zhang",
"Minxin",
""
],
[
"Liu",
"Yuxuan",
""
],
[
"Schaeffer",
"Hayden",
""
]
] |
2509.03070
|
Po-Heng Chou
|
Po-Heng Chou, Wei-Lung Mao, and Ru-Ping Lin
|
YOLO-based Bearing Fault Diagnosis With Continuous Wavelet Transform
|
5 pages, 2 figures, 2 tables, submitted to IEEE Sensors Letters
| null | null | null |
eess.SP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
This letter proposes a YOLO-based framework for spatial bearing fault diagnosis using time-frequency spectrograms derived from continuous wavelet transform (CWT). One-dimensional vibration signals are first transformed into time-frequency spectrograms using Morlet wavelets to capture transient fault signatures. These spectrograms are then processed by YOLOv9, v10, and v11 models to classify fault types. Evaluated on three benchmark datasets, including Case Western Reserve University (CWRU), Paderborn University (PU), and Intelligent Maintenance System (IMS), the proposed CWT-YOLO pipeline achieves significantly higher accuracy and generalizability than the baseline MCNN-LSTM model. Notably, YOLOv11 reaches mAP scores of 99.4% (CWRU), 97.8% (PU), and 99.5% (IMS). In addition, its region-aware detection mechanism enables direct visualization of fault locations in spectrograms, offering a practical solution for condition monitoring in rotating machinery.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 07:08:44 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 14:37:04 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Chou",
"Po-Heng",
""
],
[
"Mao",
"Wei-Lung",
""
],
[
"Lin",
"Ru-Ping",
""
]
] |
2509.03119
|
Yash Vyas
|
Yash Vyas and Matteo Bottin
|
Forbal: Force Balanced 2-5 Degree of Freedom Robot Manipulator Built from a Five Bar Linkage
| null | null | null | null |
cs.RO cs.SY eess.SY
|
http://creativecommons.org/licenses/by/4.0/
|
A force balanced manipulator design based on the closed chain planar five bar linkage is developed and experimentally validated. We present 2 variants as a modular design: Forbal-2, a planar 2-DOF manipulator, and its extension to 5-DOF spatial motion called Forbal-5. The design considerations in terms of geometric, kinematic, and dynamic design that fulfill the force balance conditions while maximizing workspace are discussed. Then, the inverse kinematics of both variants are derived from geometric principles. We validate the improvements from force balancing the manipulator through comparative experiments with counter mass balanced and unbalanced configurations. The results show how the balanced configuration yields a reduction in the average reaction moments of up to 66%, a reduction of average joint torques of up to 79%, as well as a noticeable reduction in position error for Forbal-2. For Forbal-5, which has a higher end effector payload mass, the joint torques are reduced up to 84% for the balanced configuration. Experimental results validate that the balanced manipulator design is suitable for applications where the reduction of joint torques and reaction forces/moments helps achieve millimeter level precision.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 08:20:55 GMT"
},
{
"version": "v2",
"created": "Sun, 07 Sep 2025 20:49:09 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Vyas",
"Yash",
""
],
[
"Bottin",
"Matteo",
""
]
] |
2509.03197
|
Martin \v{C}ech
|
Martin \v{C}ech
|
First moment of quadratic Dirichlet $L$-functions with secondary terms
|
Corrected title and other minor fixes; 34 pages
| null | null | null |
math.NT
|
http://creativecommons.org/licenses/by-nc-sa/4.0/
|
We study the first moment of primitive quadratic Dirichlet $L$-functions. Assuming the Riemann hypothesis and the generalized Lindel\"of hypothesis, we obtain an asymptotic formula at the central point with error $O(X^{1/4+\epsilon})$, and identify lower-order terms of size $X^{1/3}\log X$.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 10:33:50 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 15:00:45 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Čech",
"Martin",
""
]
] |
2509.03246
|
Yuchen Liao
|
Yuchen Liao and Zhipeng Liu
|
Multipoint distributions of the KPZ fixed point with compactly supported initial conditions
|
37 pages, 3 figures. Revised version, some typos and minor mistakes fixed, more references added, abstract modified
| null | null | null |
math.PR math-ph math.MP
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
The KPZ fixed point is a universal limiting space-time random field for the Kardar-Parisi-Zhang universality class. While the joint law of the KPZ fixed point at a fixed time has been studied extensively, the multipoint distributions of the KPZ fixed point in the general space-time plane are much less well understood. More explicitly, formulas were only available for the narrow wedge initial condition arXiv:1906.01053, arXiv:1907.09876 and the flat initial condition arXiv:1907.09876 for the multipoint distributions, and the half-Brownian and Brownian initial conditions arXiv:2010.07357v1, arXiv:2504.19975 for the two-point distributions.
In this paper, we obtain the first formula for the space-time joint distributions of the KPZ fixed point with general initial conditions of compact support. The formula is obtained through taking $1:2:3$ KPZ scaling limit of the multipoint distribution formulas for the totally asymmetric simple exclusion process (TASEP). A key ingredient is a probabilistic representation, inspired by arXiv:1701.00018, of the kernel encoding the initial condition for TASEP, which was first defined through an implicit characterization in arXiv:1907.09876. Moreover, we also verify that the equal time version of our formula matches the path integral formula in arXiv:1701.00018 for the KPZ fixed point.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 12:04:45 GMT"
},
{
"version": "v2",
"created": "Thu, 04 Sep 2025 16:46:49 GMT"
},
{
"version": "v3",
"created": "Mon, 08 Sep 2025 16:20:21 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Liao",
"Yuchen",
""
],
[
"Liu",
"Zhipeng",
""
]
] |
2509.03333
|
Tianfu Qi
|
Tianfu Qi, Jun Wang
|
Cutoff Rate Bounds and Constellation Shaping under Mixed Gaussian-Impulsive Noise
| null | null | null | null |
eess.SP cs.IT math.IT
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Mixed noise, composed of white Gaussian noise (WGN) and impulsive noise (IN), appears in numerous communication scenarios and can severely degrade system performance. In this paper, we optimize the transmitted constellation under mixed noise based on a theoretical analysis of the cutoff rate (CR). First, starting from the passband model of the mixed noise, we derive its corresponding baseband representation. Then, the baseband model is employed to obtain closed-form lower and upper bounds of the CR. A piecewise linear approximation is applied to derive efficient bounds by exploiting the algebraic properties of the integral terms. These bounds are then used as criteria to optimize the transmitted constellation points in both geometric positions and probabilistic distributions. The projected gradient method is employed to solve the optimization problem, and the convergence and properties of the solutions are analyzed. Numerical results demonstrate that the proposed CR bounds are tight and exhibit the expected asymptotic behavior. Furthermore, the optimized constellation scheme achieves a significant rate improvement compared to baselines.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 14:09:44 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 12:11:32 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Qi",
"Tianfu",
""
],
[
"Wang",
"Jun",
""
]
] |
2509.03377
|
Rui Xie
|
Rui Xie, Asad Ul Haq, Linsen Ma, Yunhua Fang, Zirak Burzin Engineer, Liu Liu, Tong Zhang
|
Amplifying Effective CXL Memory Bandwidth for LLM Inference via Transparent Near-Data Processing
| null | null | null | null |
cs.AR
|
http://arxiv.org/licenses/nonexclusive-distrib/1.0/
|
Large language model (LLM) inference is bottlenecked by the limited bandwidth of CXL-based memory used for capacity expansion. We introduce CXL-NDP, a transparent near-data processing architecture that amplifies effective CXL bandwidth without requiring changes to the CXL.mem interface or AI models. CXL-NDP integrates a precision-scalable bit-plane layout for dynamic quantization with transparent lossless compression of weights and KV caches directly within the CXL device. In end-to-end serving, CXL-NDP improves throughput by 43%, extends the maximum context length by 87%, and reduces the KV cache footprint by 46.9% without accuracy loss. Hardware synthesis confirms its practicality with a modest silicon footprint, lowering the barrier for adopting efficient, scalable CXL-based memory in generative AI infrastructure.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 14:53:45 GMT"
},
{
"version": "v2",
"created": "Mon, 08 Sep 2025 17:22:17 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Xie",
"Rui",
""
],
[
"Haq",
"Asad Ul",
""
],
[
"Ma",
"Linsen",
""
],
[
"Fang",
"Yunhua",
""
],
[
"Engineer",
"Zirak Burzin",
""
],
[
"Liu",
"Liu",
""
],
[
"Zhang",
"Tong",
""
]
] |
2509.03494
|
Mohammed El Hassouni
|
Yahya Benmahane, Mohammed El Hassouni
|
Parameter-Efficient Adaptation of mPLUG-Owl2 via Pixel-Level Visual Prompts for NR-IQA
| null | null | null | null |
cs.CV
|
http://creativecommons.org/licenses/by/4.0/
|
In this paper, we propose a novel parameter-efficient adaptation method for No- Reference Image Quality Assessment (NR-IQA) using visual prompts optimized in pixel-space. Unlike full fine-tuning of Multimodal Large Language Models (MLLMs), our approach trains only 600K parameters at most (< 0.01% of the base model), while keeping the underlying model fully frozen. During inference, these visual prompts are combined with images via addition and processed by mPLUG-Owl2 with the textual query "Rate the technical quality of the image." Evaluations across distortion types (synthetic, realistic, AI-generated) on KADID- 10k, KonIQ-10k, and AGIQA-3k demonstrate competitive performance against full finetuned methods and specialized NR-IQA models, achieving 0.93 SRCC on KADID-10k. To our knowledge, this is the first work to leverage pixel-space visual prompts for NR-IQA, enabling efficient MLLM adaptation for low-level vision tasks. The source code is publicly available at https: // github. com/ yahya-ben/ mplug2-vp-for-nriqa.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 17:23:24 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 16:31:33 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Benmahane",
"Yahya",
""
],
[
"Hassouni",
"Mohammed El",
""
]
] |
2509.03558
|
Wenbo Gu GUwenbo
|
Wenbo Gu, Ali Esamdin, Cunhai Bai, Jicheng Zhang, Xuan Zhang, Guojie Feng, Letian Wang, Guangxin Pu, Xinliang Wang, Daiping Zhang
|
Seeing and Meteorological Analysis at the North-1 and North-2 Points of Muztagh-Ata Site on the Pamir Plateau
|
14 pages,26 figures. published in MNRAS
| null |
10.1093/mnras/staf1210
| null |
astro-ph.IM
|
http://creativecommons.org/licenses/by-nc-nd/4.0/
|
To support the selection of large optical/infrared telescope sites in western China, long-term monitoring of atmospheric conditions and astronomical seeing has been conducted at the Muztagh-Ata site on the Pamir Plateau since 2017. With the monitoring focus gradually shifting northward, three stations were established: the South Point, North-1 point, and North-2 point. The North-1 point,selected as the site for the Muztagh-Ata 1.93 m Synergy Telescope (MOST), has recorded seeing and meteorological parameters since late 2018. In 2023,the North-2 point was established approximately 1.5 km northeast of North-1 point as a candidate location for a future large-aperture telescope. A 10m DIMM tower and a PC-4A environmental monitoring system were installed to evaluate site quality. This study presents a comparative analysis of data from the North-1 and North-2 points during 2018-2024.The median seeing is 0.89 arcsecs at North-1 and 0.78 arcsecs at North-2. Both points show clear seasonal and diurnal variations,with winter nights offering optimal observing conditions.On average, about 64% of the nighttime duration per year is suitable for astronomical observations. Nighttime temperature variation is low :2.03 at North-1 and 2.10 at North-2 .Median wind speeds are 5-6 m/s, with dominant directions between 210 and 300, contributing to stable airflow. Moderate wind suppresses turbulence, while strong shear and rapid fluctuations degrade image quality. These findings confirm that both the North-1 and North-2 points offer high-quality atmospheric conditions and serve as promising sites for future ground-based optical/infrared telescopes in western China.
|
[
{
"version": "v1",
"created": "Wed, 03 Sep 2025 09:15:21 GMT"
},
{
"version": "v2",
"created": "Sat, 06 Sep 2025 17:42:37 GMT"
}
] | 2025-09-09T00:00:00 |
[
[
"Gu",
"Wenbo",
""
],
[
"Esamdin",
"Ali",
""
],
[
"Bai",
"Cunhai",
""
],
[
"Zhang",
"Jicheng",
""
],
[
"Zhang",
"Xuan",
""
],
[
"Feng",
"Guojie",
""
],
[
"Wang",
"Letian",
""
],
[
"Pu",
"Guangxin",
""
],
[
"Wang",
"Xinliang",
""
],
[
"Zhang",
"Daiping",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.