# Research Papers — 2025-11 · bundle page 180 of 197

> Full-text markdown of papers #8951–9000 of 9830 published in 2025-11, newest first. AI-readable mirrors of the full 9830-paper 2025-11 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2025-11/page/180
[← Previous bundle](https://aipod.dev/papers/archive/2025-11/page/179.md)
[Next bundle →](https://aipod.dev/papers/archive/2025-11/page/181.md)

### DANIEL: A Distributed and Scalable Approach for Global Representation Learning with EHR Applications

Authors: Zebin Wang, Ziming Gan, Weijing Tang, Zongqi Xia, Tianrun Cai, Tianxi Cai, Junwei Lu
Published: 2025-11-04
Categories: stat.ME, cs.LG
arXiv: https://arxiv.org/abs/2511.02754v1

Classical probabilistic graphical models face fundamental challenges in modern data environments, which are characterized by high dimensionality, source heterogeneity, and stringent data-sharing constraints. In this work, we revisit the Ising model, a well-established member of the Markov Random Field (MRF) family, and develop a distributed framework that enables scalable and privacy-preserving representation learning from large-scale binary data with inherent low-rank structure. Our approach optimizes a non-convex surrogate loss function via bi-factored gradient descent, offering substantial computational and communication advantages over conventional convex approaches. We evaluate our algorithm on multi-institutional electronic health record (EHR) datasets from 58,248 patients across the University of Pittsburgh Medical Center (UPMC) and Mass General Brigham (MGB), demonstrating superior performance in global representation learning and downstream clinical tasks, including relationship detection, patient phenotyping, and patient clustering. These results highlight a broader potential for statistical inference in federated, high-dimensional settings while addressing the practical challenges of data complexity and multi-institutional integration.

---

### AI Diffusion in Low Resource Language Countries

Authors: Amit Misra, Syed Waqas Zamir, Wassim Hamidouche, Inbal Becker-Reshef, Juan Lavista Ferres
Published: 2025-11-04
Categories: cs.CL, cs.AI, cs.CY
arXiv: https://arxiv.org/abs/2511.02752v1

Artificial intelligence (AI) is diffusing globally at unprecedented speed, but adoption remains uneven. Frontier Large Language Models (LLMs) are known to perform poorly on low-resource languages due to data scarcity. We hypothesize that this performance deficit reduces the utility of AI, thereby slowing adoption in Low-Resource Language Countries (LRLCs). To test this, we use a weighted regression model to isolate the language effect from socioeconomic and demographic factors, finding that LRLCs have a share of AI users that is approximately 20% lower relative to their baseline. These results indicate that linguistic accessibility is a significant, independent barrier to equitable AI diffusion.

---

### Using Span Queries to Optimize for Cache and Attention Locality

Authors: Paul Castro, Nick Mitchell, Nathan Ordonez, Thomas Parnell, Mudhakar Srivatsa, Antoni Viros i Martin
Published: 2025-11-04
Categories: cs.AI
arXiv: https://arxiv.org/abs/2511.02749v1

Clients are evolving beyond chat completion, and now include a variety of innovative inference-time scaling and deep reasoning techniques. At the same time, inference servers remain heavily optimized for chat completion. Prior work has shown that large improvements to KV cache hit rate are possible if inference servers evolve towards these non-chat use cases. However, they offer solutions that are also optimized for a single use case, RAG. In this paper, we introduce the span query to generalize the interface to the inference server. We demonstrate that chat, RAG, inference-time scaling, and agentic workloads can all be expressed as span queries. We show how the critical distinction that had been assumed by prior work lies in whether the order of the inputs matter -- do they commute? In chat, they do not. In RAG, they often do. This paper introduces span queries, which are expression trees of inference calls, linked together with commutativity constraints. We describe span query syntax and semantics. We show how they can be automatically optimized to improve KV cache locality. We show how a small change to vLLM (affecting only 492 lines) can enable high-performance execution of span queries. Using this stack, we demonstrate that span queries can achieve 10-20x reductions in TTFT for two distinct non-chat use cases. Finally, we show that span queries can also be optimized to improve attention locality, so as to avoid the so-called lost-in-the-middle problem. We demonstrate that an attention-optimized span query on a 2b parameter model vastly outperforms the accuracy of a stock inference server using an 8b model.

---

### Agentic World Modeling for 6G: Near-Real-Time Generative State-Space Reasoning

Authors: Farhad Rezazadeh, Amir Ashtari Gargari, Hatim Chergui, Sandra Lagen, Merouane Debbah, Houbing Song, Lingjia Liu
Published: 2025-11-04
Categories: cs.NI, cs.LG
arXiv: https://arxiv.org/abs/2511.02748v2

We argue that sixth-generation (6G) intelligence is not fluent token prediction but the capacity to imagine and choose -- to simulate future scenarios, weigh trade-offs, and act with calibrated uncertainty. We reframe open radio access network (O-RAN) near-real-time (Near-RT) control via counterfactual dynamics and a world modeling (WM) paradigm that learns an action-conditioned generative state space. This enables quantitative "what-if" forecasting beyond large language models (LLMs) as the primary modeling primitive. Actions such as physical resource blocks (PRBs) are treated as first-class control inputs in a causal world model, and both aleatoric and epistemic uncertainty are modeled for prediction and what-if analysis. An agentic, model predictive control (MPC)-based cross-entropy method (CEM) planner operates over short horizons, using prior-mean rollouts within data-driven PRB bounds to maximize a deterministic reward. The model couples multi-scale structured state-space mixtures (MS3M) with a compact stochastic latent to form WM-MS3M, summarizing key performance indicators (KPIs) histories and predicting next-step KPIs under hypothetical PRB sequences. On realistic O-RAN traces, WM-MS3M cuts mean absolute error (MAE) by 1.69% versus MS3M with 32% fewer parameters and similar latency, and achieves 35-80% lower root mean squared error (RMSE) than attention/hybrid baselines with 2.3-4.1x faster inference, enabling rare-event simulation and offline policy screening.

---

### Agentic World Modeling for 6G: Near-Real-Time Generative State-Space Reasoning

Authors: Farhad Rezazadeh, Hatim Chergui, Merouane Debbah, Houbing Song, Dusit Niyato, Lingjia Liu
Published: 2025-11-04
Categories: cs.NI, cs.LG
arXiv: https://arxiv.org/abs/2511.02748v1

We argue that sixth-generation (6G) intelligence is not fluent token prediction but the capacity to imagine and choose -- to simulate future scenarios, weigh trade-offs, and act with calibrated uncertainty. We reframe open radio access network (O-RAN) near-real-time (Near-RT) control via counterfactual dynamics and a world modeling (WM) paradigm that learns an action-conditioned generative state space. This enables quantitative "what-if" forecasting beyond large language models (LLMs) as the primary modeling primitive. Actions such as physical resource blocks (PRBs) are treated as first-class control inputs in a causal world model, and both aleatoric and epistemic uncertainty are modeled for prediction and what-if analysis. An agentic, model predictive control (MPC)-based cross-entropy method (CEM) planner operates over short horizons, using prior-mean rollouts within data-driven PRB bounds to maximize a deterministic reward. The model couples multi-scale structured state-space mixtures (MS3M) with a compact stochastic latent to form WM-MS3M, summarizing key performance indicators (KPIs) histories and predicting next-step KPIs under hypothetical PRB sequences. On realistic O-RAN traces, WM-MS3M cuts mean absolute error (MAE) by 1.69% versus MS3M with 32% fewer parameters and similar latency, and achieves 35-80% lower root mean squared error (RMSE) than attention/hybrid baselines with 2.3-4.1x faster inference, enabling rare-event simulation and offline policy screening.

---

### Calibration improves detection of mislabeled examples

Authors: Ilies Chibane, Thomas George, Pierre Nodet, Vincent Lemaire
Published: 2025-11-04
Categories: cs.LG
arXiv: https://arxiv.org/abs/2511.02738v1

Mislabeled data is a pervasive issue that undermines the performance of machine learning systems in real-world applications. An effective approach to mitigate this problem is to detect mislabeled instances and subject them to special treatment, such as filtering or relabeling. Automatic mislabeling detection methods typically rely on training a base machine learning model and then probing it for each instance to obtain a trust score that each provided label is genuine or incorrect. The properties of this base model are thus of paramount importance. In this paper, we investigate the impact of calibrating this model. Our empirical results show that using calibration methods improves the accuracy and robustness of mislabeled instance detection, providing a practical and effective solution for industrial applications.

---

### CostBench: Evaluating Multi-Turn Cost-Optimal Planning and Adaptation in Dynamic Environments for LLM Tool-Use Agents

Authors: Jiayu Liu, Cheng Qian, Zhaochen Su, Qing Zong, Shijue Huang, Bingxiang He, Yi R. Fung
Published: 2025-11-04
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2511.02734v3

Current evaluations of Large Language Model (LLM) agents primarily emphasize task completion, often overlooking resource efficiency and adaptability. This neglects a crucial capability: agents' ability to devise and adjust cost-optimal plans in response to changing environments. To bridge this gap, we introduce CostBench, a scalable, cost-centric benchmark designed to evaluate agents' economic reasoning and replanning abilities. Situated in the travel-planning domain, CostBench comprises tasks solvable via multiple sequences of atomic and composite tools with diverse, customizable costs. It also supports four types of dynamic blocking events, such as tool failures and cost changes, to simulate real-world unpredictability and necessitate agents to adapt in real time. Evaluating leading open-sourced and proprietary models on CostBench reveals a substantial gap in cost-aware planning: agents frequently fail to identify cost-optimal solutions in static settings, with even GPT-5 achieving less than 75% exact match rate on the hardest tasks, and performance further dropping by around 40% under dynamic conditions. By diagnosing these weaknesses, CostBench lays the groundwork for developing future agents that are both economically rational and robust.

---

### CostBench: Evaluating Multi-Turn Cost-Optimal Planning and Adaptation in Dynamic Environments for LLM Tool-Use Agents

Authors: Jiayu Liu, Cheng Qian, Zhaochen Su, Qing Zong, Shijue Huang, Bingxiang He, Yi R. Fung
Published: 2025-11-04
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2511.02734v1

Current evaluations of Large Language Model (LLM) agents primarily emphasize task completion, often overlooking resource efficiency and adaptability. This neglects a crucial capability: agents' ability to devise and adjust cost-optimal plans in response to changing environments. To bridge this gap, we introduce CostBench, a scalable, cost-centric benchmark designed to evaluate agents' economic reasoning and replanning abilities. Situated in the travel-planning domain, CostBench comprises tasks solvable via multiple sequences of atomic and composite tools with diverse, customizable costs. It also supports four types of dynamic blocking events, such as tool failures and cost changes, to simulate real-world unpredictability and necessitate agents to adapt in real time. Evaluating leading open-sourced and proprietary models on CostBench reveals a substantial gap in cost-aware planning: agents frequently fail to identify cost-optimal solutions in static settings, with even GPT-5 achieving less than 75% exact match rate on the hardest tasks, and performance further dropping by around 40% under dynamic conditions. By diagnosing these weaknesses, CostBench lays the groundwork for developing future agents that are both economically rational and robust.

---

### Association-sensory spatiotemporal hierarchy and functional gradient-regularised recurrent neural network with implications for schizophrenia

Authors: Subati Abulikemu, Puria Radmard, Michail Mamalakis, John Suckling
Published: 2025-11-04
Categories: q-bio.NC, cs.LG
arXiv: https://arxiv.org/abs/2511.02722v1

The human neocortex is functionally organised at its highest level along a continuous sensory-to-association (AS) hierarchy. This study characterises the AS hierarchy of patients with schizophrenia in a comparison with controls. Using a large fMRI dataset (N=355), we extracted individual AS gradients via spectral analysis of brain connectivity, quantified hierarchical specialisation by gradient spread, and related this spread with connectivity geometry. We found that schizophrenia compresses the AS hierarchy indicating reduced functional differentiation. By modelling neural timescale with the Ornstein-Uhlenbeck process, we observed that the most specialised, locally cohesive regions at the gradient extremes exhibit dynamics with a longer time constant, an effect that is attenuated in schizophrenia. To study computation, we used the gradients to regularise subject-specific recurrent neural networks (RNNs) trained on working memory tasks. Networks endowed with greater gradient spread learned more efficiently, plateaued at lower task loss, and maintained stronger alignment to the prescribed AS hierarchical geometry. Fixed point linearisation showed that high-range networks settled into more stable neural states during memory delay, evidenced by lower energy and smaller maximal Jacobian eigenvalues. This gradient-regularised RNN framework therefore links large-scale cortical architecture with fixed point stability, providing a mechanistic account of how gradient de-differentiation could destabilise neural computations in schizophrenia, convergently supported by empirical timescale flattening and model-based evidence of less stable fixed points.

---

### PETra: A Multilingual Corpus of Pragmatic Explicitation in Translation

Authors: Doreen Osmelak, Koel Dutta Chowdhury, Uliana Sentsova, Cristina España-Bonet, Josef van Genabith
Published: 2025-11-04
Categories: cs.CL
arXiv: https://arxiv.org/abs/2511.02721v2

Translators often enrich texts with background details that make implicit cultural meanings explicit for new audiences. This phenomenon, known as pragmatic explicitation, has been widely discussed in translation theory but rarely modeled computationally. We introduce PragExTra, the first multilingual corpus and detection framework for pragmatic explicitation. The corpus covers eight language pairs from TED-Multi and Europarl and includes additions such as entity descriptions, measurement conversions, and translator remarks. We identify candidate explicitation cases through null alignments and refined using active learning with human annotation. Our results show that entity and system-level explicitations are most frequent, and that active learning improves classifier accuracy by 7-8 percentage points, achieving up to 0.88 accuracy and 0.82 F1 across languages. PragExTra establishes pragmatic explicitation as a measurable, cross-linguistic phenomenon and takes a step towards building culturally aware machine translation. Keywords: translation, multilingualism, explicitation

---

### PragExTra: A Multilingual Corpus of Pragmatic Explicitation in Translation

Authors: Doreen Osmelak, Koel Dutta Chowdhury, Uliana Sentsova, Cristina España-Bonet, Josef van Genabith
Published: 2025-11-04
Categories: cs.CL
arXiv: https://arxiv.org/abs/2511.02721v1

Translators often enrich texts with background details that make implicit cultural meanings explicit for new audiences. This phenomenon, known as pragmatic explicitation, has been widely discussed in translation theory but rarely modeled computationally. We introduce PragExTra, the first multilingual corpus and detection framework for pragmatic explicitation. The corpus covers eight language pairs from TED-Multi and Europarl and includes additions such as entity descriptions, measurement conversions, and translator remarks. We identify candidate explicitation cases through null alignments and refined using active learning with human annotation. Our results show that entity and system-level explicitations are most frequent, and that active learning improves classifier accuracy by 7-8 percentage points, achieving up to 0.88 accuracy and 0.82 F1 across languages. PragExTra establishes pragmatic explicitation as a measurable, cross-linguistic phenomenon and takes a step towards building culturally aware machine translation. Keywords: translation, multilingualism, explicitation

---

### LLEXICORP: End-user Explainability of Convolutional Neural Networks

Authors: Vojtěch Kůr, Adam Bajger, Adam Kukučka, Marek Hradil, Vít Musil, Tomáš Brázdil
Published: 2025-11-04
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2511.02720v1

Convolutional neural networks (CNNs) underpin many modern computer vision systems. With applications ranging from common to critical areas, a need to explain and understand the model and its decisions (XAI) emerged. Prior works suggest that in the top layers of CNNs, the individual channels can be attributed to classifying human-understandable concepts. Concept relevance propagation (CRP) methods can backtrack predictions to these channels and find images that most activate these channels. However, current CRP workflows are largely manual: experts must inspect activation images to name the discovered concepts and must synthesize verbose explanations from relevance maps, limiting the accessibility of the explanations and their scalability. To address these issues, we introduce Large Language model EXplaIns COncept Relevance Propagation (LLEXICORP), a modular pipeline that couples CRP with a multimodal large language model. Our approach automatically assigns descriptive names to concept prototypes and generates natural-language explanations that translate quantitative relevance distributions into intuitive narratives. To ensure faithfulness, we craft prompts that teach the language model the semantics of CRP through examples and enforce a separation between naming and explanation tasks. The resulting text can be tailored to different audiences, offering low-level technical descriptions for experts and high-level summaries for non-technical stakeholders. We qualitatively evaluate our method on various images from ImageNet on a VGG16 model. Our findings suggest that integrating concept-based attribution methods with large language models can significantly lower the barrier to interpreting deep neural networks, paving the way for more transparent AI systems.

---

### Does Interpretability of Knowledge Tracing Models Support Teacher Decision Making?

Authors: Adia Khalid, Alina Deriyeva, Benjamin Paassen
Published: 2025-11-04
Categories: cs.LG, cs.HC
arXiv: https://arxiv.org/abs/2511.02718v1

Knowledge tracing (KT) models are a crucial basis for pedagogical decision-making, namely which task to select next for a learner and when to stop teaching a particular skill. Given the high stakes of pedagogical decisions, KT models are typically required to be interpretable, in the sense that they should implement an explicit model of human learning and provide explicit estimates of learners' abilities. However, to our knowledge, no study to date has investigated whether the interpretability of KT models actually helps human teachers to make teaching decisions. We address this gap. First, we perform a simulation study to show that, indeed, decisions based on interpretable KT models achieve mastery faster compared to decisions based on a non-interpretable model. Second, we repeat the study but ask $N=12$ human teachers to make the teaching decisions based on the information provided by KT models. As expected, teachers rate interpretable KT models higher in terms of usability and trustworthiness. However, the number of tasks needed until mastery hardly differs between KT models. This suggests that the relationship between model interpretability and teacher decisions is not straightforward: teachers do not solely rely on KT models to make decisions and further research is needed to investigate how learners and teachers actually understand and use KT models.

---

### An unscented Kalman filter method for real time input-parameter-state estimation

Authors: Marios Impraimakis, Andrew W. Smyth
Published: 2025-11-04
Categories: eess.SP, cs.AI, cs.CV, eess.AS, eess.SY
arXiv: https://arxiv.org/abs/2511.02717v1

The input-parameter-state estimation capabilities of a novel unscented Kalman filter is examined herein on both linear and nonlinear systems. The unknown input is estimated in two stages within each time step. Firstly, the predicted dynamic states and the system parameters provide an estimation of the input. Secondly, the corrected with measurements states and parameters provide a final estimation. Importantly, it is demonstrated using the perturbation analysis that, a system with at least a zero or a non-zero known input can potentially be uniquely identified. This output-only methodology allows for a better understanding of the system compared to classical output-only parameter identification strategies, given that all the dynamic states, the parameters, and the input are estimated jointly and in real-time.

---

### VidEmo: Affective-Tree Reasoning for Emotion-Centric Video Foundation Models

Authors: Zhicheng Zhang, Weicheng Wang, Yongjie Zhu, Wenyu Qin, Pengfei Wan, Di Zhang, Jufeng Yang
Published: 2025-11-04
Categories: cs.CV
arXiv: https://arxiv.org/abs/2511.02712v1

Understanding and predicting emotion from videos has gathered significant attention in recent studies, driven by advancements in video large language models (VideoLLMs). While advanced methods have made progress in video emotion analysis, the intrinsic nature of emotions poses significant challenges. Emotions are characterized by dynamic and cues-dependent properties, making it difficult to understand complex and evolving emotional states with reasonable rationale. To tackle these challenges, we propose a novel affective cues-guided reasoning framework that unifies fundamental attribute perception, expression analysis, and high-level emotional understanding in a stage-wise manner. At the core of our approach is a family of video emotion foundation models (VidEmo), specifically designed for emotion reasoning and instruction-following. These models undergo a two-stage tuning process: first, curriculum emotion learning for injecting emotion knowledge, followed by affective-tree reinforcement learning for emotion reasoning. Moreover, we establish a foundational data infrastructure and introduce a emotion-centric fine-grained dataset (Emo-CFG) consisting of 2.1M diverse instruction-based samples. Emo-CFG includes explainable emotional question-answering, fine-grained captions, and associated rationales, providing essential resources for advancing emotion understanding tasks. Experimental results demonstrate that our approach achieves competitive performance, setting a new milestone across 15 face perception tasks.

---

### Optimizing Kernel Discrepancies via Subset Selection

Authors: Deyao Chen, François Clément, Carola Doerr, Nathan Kirk
Published: 2025-11-04
Categories: stat.ML, cs.CG, cs.LG, math.NA
arXiv: https://arxiv.org/abs/2511.02706v1

Kernel discrepancies are a powerful tool for analyzing worst-case errors in quasi-Monte Carlo (QMC) methods. Building on recent advances in optimizing such discrepancy measures, we extend the subset selection problem to the setting of kernel discrepancies, selecting an m-element subset from a large population of size $n \gg m$. We introduce a novel subset selection algorithm applicable to general kernel discrepancies to efficiently generate low-discrepancy samples from both the uniform distribution on the unit hypercube, the traditional setting of classical QMC, and from more general distributions $F$ with known density functions by employing the kernel Stein discrepancy. We also explore the relationship between the classical $L_2$ star discrepancy and its $L_\infty$ counterpart.

---

### Curriculum Design for Trajectory-Constrained Agent: Compressing Chain-of-Thought Tokens in LLMs

Authors: Georgios Tzannetos, Parameswaran Kamalaruban, Adish Singla
Published: 2025-11-04
Categories: cs.LG
arXiv: https://arxiv.org/abs/2511.02690v1

Training agents to operate under strict constraints during deployment, such as limited resource budgets or stringent safety requirements, presents significant challenges, especially when these constraints render the task complex. In this work, we propose a curriculum learning strategy that gradually tightens constraints during training, enabling the agent to incrementally master the deployment requirements. Inspired by self-paced learning techniques in unconstrained reinforcement learning (RL), our approach facilitates a smoother transition to challenging environments by initially training on simplified versions of the constraints and progressively introducing the full deployment conditions. We provide a theoretical analysis using an RL agent in a binary-tree Markov Decision Process (MDP) to demonstrate that our curriculum strategy can accelerate training relative to a baseline approach that imposes the trajectory constraints from the outset. Moreover, we empirically validate the effectiveness and generality of our method across both RL and large language model (LLM) agents in diverse settings, including a binary-tree MDP, a multi-task navigation domain, and a math reasoning task with two benchmarks. These results highlight the potential of curriculum design in enhancing the efficiency and performance of agents operating under complex trajectory constraints during deployment. Moreover, when applied to LLMs, our strategy enables compression of output chain-of-thought tokens, achieving a substantial inference speedup on consumer hardware, demonstrating its effectiveness for resource-constrained deployment.

---

### The Collaboration Gap: Exploration and Benchmarking of Open-World Agentic Cooperation

Authors: Tim R. Davidson, Adam Fourney, Saleema Amershi, Robert West, Eric Horvitz, Ece Kamar
Published: 2025-11-04
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2511.02687v2

The trajectory of AI development suggests that we will increasingly rely on agent-based systems powered by language models, composed of independently developed agents with different information, privileges, and tools. The success of these systems will depend critically on effective collaboration among these heterogeneous agents, even under partial observability. Despite intense interest, the literature lacks empirical studies evaluating agentic collaboration without relying on fixed communication protocols, limiting insights for open-world deployments. We propose an illustrative collaborative maze-solving benchmark that (i) isolates collaborative capabilities, (ii) modulates problem complexity, (iii) enables scalable automated grading, and (iv) imposes no output-format constraints, benchmarking unguided, natural communication. Using this benchmark, we evaluate 32 leading open- and closed-source models in solo, homogeneous, and heterogeneous pairings. Our results reveal a surprising collaboration gap: models that perform well solo often degrade substantially when required to collaborate. We identify mitigations that show remarkable influence. For example, a small nudge via a relay inference approach, where a stronger agent leads before handing off to a weaker one, closes much of the gap. Our findings argue for (1) collaboration-aware evaluation, (2) training strategies to enhance collaborative capabilities, and (3) deliberate interaction design to elicit agents' collaboration skills, principles relevant to AI-AI and human-AI settings where agents must establish common ground.

---

### The Collaboration Gap

Authors: Tim R. Davidson, Adam Fourney, Saleema Amershi, Robert West, Eric Horvitz, Ece Kamar
Published: 2025-11-04
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2511.02687v1

The trajectory of AI development suggests that we will increasingly rely on agent-based systems composed of independently developed agents with different information, privileges, and tools. The success of these systems will critically depend on effective collaboration among these heterogeneous agents, even under partial observability. Despite intense interest, few empirical studies have evaluated such agent-agent collaboration at scale. We propose a collaborative maze-solving benchmark that (i) isolates collaborative capabilities, (ii) modulates problem complexity, (iii) enables scalable automated grading, and (iv) imposes no output-format constraints, preserving ecological plausibility. Using this framework, we evaluate 32 leading open- and closed-source models in solo, homogeneous, and heterogeneous pairings. Our results reveal a "collaboration gap": models that perform well solo often degrade substantially when required to collaborate. Collaboration can break down dramatically; for instance, small distilled models that solve mazes well alone may fail almost completely in certain pairings. We find that starting with the stronger agent often improves outcomes, motivating a "relay inference" approach where the stronger agent leads before handing off to the weaker one, closing much of the gap. Our findings argue for (1) collaboration-aware evaluation, (2) training strategies developed to enhance collaborative capabilities, and (3) interaction design that reliably elicits agents' latent skills, guidance that applies to AI-AI and human-AI collaboration.

---

### Modality-Transition Representation Learning for Visible-Infrared Person Re-Identification

Authors: Chao Yuan, Zanwu Liu, Guiwei Zhang, Haoxuan Xu, Yujian Zhao, Guanglin Niu, Bo Li
Published: 2025-11-04
Categories: cs.CV
arXiv: https://arxiv.org/abs/2511.02685v1

Visible-infrared person re-identification (VI-ReID) technique could associate the pedestrian images across visible and infrared modalities in the practical scenarios of background illumination changes. However, a substantial gap inherently exists between these two modalities. Besides, existing methods primarily rely on intermediate representations to align cross-modal features of the same person. The intermediate feature representations are usually create by generating intermediate images (kind of data enhancement), or fusing intermediate features (more parameters, lack of interpretability), and they do not make good use of the intermediate features. Thus, we propose a novel VI-ReID framework via Modality-Transition Representation Learning (MTRL) with a middle generated image as a transmitter from visible to infrared modals, which are fully aligned with the original visible images and similar to the infrared modality. After that, using a modality-transition contrastive loss and a modality-query regularization loss for training, which could align the cross-modal features more effectively. Notably, our proposed framework does not need any additional parameters, which achieves the same inference speed to the backbone while improving its performance on VI-ReID task. Extensive experimental results illustrate that our model significantly and consistently outperforms existing SOTAs on three typical VI-ReID datasets.

---

### Optimal Singular Damage: Efficient LLM Inference in Low Storage Regimes

Authors: Mohammadsajad Alipour, Mohammad Mohammadi Amiri
Published: 2025-11-04
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2511.02681v1

Large language models (LLMs) are increasingly prevalent across diverse applications. However, their enormous size limits storage and processing capabilities to a few well-resourced stakeholders. As a result, most applications rely on pre-trained LLMs, fine-tuned for specific tasks. However, even storing the fine-tuned versions of these models remains a significant challenge due to the wide range of tasks they address. Recently, studies show that fine-tuning these models primarily affects a small fraction of parameters, highlighting the need for more efficient storage of fine-tuned models. This paper focuses on efficient storage of parameter updates in pre-trained models after fine-tuning. To address this challenge, we leverage the observation that fine-tuning updates are both low-rank and sparse, which can be utilized for storage efficiency. However, using only low-rank approximation or sparsification may discard critical singular components that enhance model expressivity. We first observe that given the same memory budget, sparsified low-rank approximations with larger ranks outperform standard low-rank approximations with smaller ranks. Building on this, we propose our method, optimal singular damage, that selectively sparsifies low-rank approximated updates by leveraging the interleaved importance of singular vectors, ensuring that the most impactful components are retained. We demonstrate through extensive experiments that our proposed methods lead to significant storage efficiency and superior accuracy within the same memory budget compared to employing the low-rank approximation or sparsification individually.

---

### RL-Aided Cognitive ISAC: Robust Detection and Sensing-Communication Trade-offs

Authors: Adam Umra, Aya M. Ahmed, Aydin Sezgin
Published: 2025-11-04
Categories: eess.SP, cs.LG
arXiv: https://arxiv.org/abs/2511.02672v1

This paper proposes a reinforcement learning (RL)-aided cognitive framework for massive MIMO-based integrated sensing and communication (ISAC) systems employing a uniform planar array (UPA). The focus is on enhancing radar sensing performance in environments with unknown and dynamic disturbance characteristics. A Wald-type detector is employed for robust target detection under non-Gaussian clutter, while a SARSA-based RL algorithm enables adaptive estimation of target positions without prior environmental knowledge. Based on the RL-derived sensing information, a joint waveform optimization strategy is formulated to balance radar sensing accuracy and downlink communication throughput. The resulting design provides an adaptive trade-off between detection performance and achievable sum rate through an analytically derived closed-form solution. Monte Carlo simulations demonstrate that the proposed cognitive ISAC framework achieves significantly improved detection probability compared to orthogonal and non-learning adaptive baselines, while maintaining competitive communication performance. These results underline the potential of RL-assisted sensing for robust and spectrum-efficient ISAC in next-generation wireless networks.

---

### Scalable Evaluation and Neural Models for Compositional Generalization

Authors: Giacomo Camposampiero, Pietro Barbiero, Michael Hersche, Roger Wattenhofer, Abbas Rahimi
Published: 2025-11-04
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2511.02667v2

Compositional generalization-a key open challenge in modern machine learning-requires models to predict unknown combinations of known concepts. However, assessing compositional generalization remains a fundamental challenge due to the lack of standardized evaluation protocols and the limitations of current benchmarks, which often favor efficiency over rigor. At the same time, general-purpose vision architectures lack the necessary inductive biases, and existing approaches to endow them compromise scalability. As a remedy, this paper introduces: 1) a rigorous evaluation framework that unifies and extends previous approaches while reducing computational requirements from combinatorial to constant; 2) an extensive and modern evaluation on the status of compositional generalization in supervised vision backbones, training more than 5000 models; 3) Attribute Invariant Networks, a class of models establishing a new Pareto frontier in compositional generalization, achieving a 23.43% accuracy improvement over baselines while reducing parameter overhead from 600% to 16% compared to fully disentangled counterparts. Our code is available at https://github.com/IBM/scalable-compositional-generalization.

---

### Scalable Evaluation and Neural Models for Compositional Generalization

Authors: Giacomo Camposampiero, Pietro Barbiero, Michael Hersche, Roger Wattenhofer, Abbas Rahimi
Published: 2025-11-04
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2511.02667v1

Compositional generalization-a key open challenge in modern machine learning-requires models to predict unknown combinations of known concepts. However, assessing compositional generalization remains a fundamental challenge due to the lack of standardized evaluation protocols and the limitations of current benchmarks, which often favor efficiency over rigor. At the same time, general-purpose vision architectures lack the necessary inductive biases, and existing approaches to endow them compromise scalability. As a remedy, this paper introduces: 1) a rigorous evaluation framework that unifies and extends previous approaches while reducing computational requirements from combinatorial to constant; 2) an extensive and modern evaluation on the status of compositional generalization in supervised vision backbones, training more than 5000 models; 3) Attribute Invariant Networks, a class of models establishing a new Pareto frontier in compositional generalization, achieving a 23.43% accuracy improvement over baselines while reducing parameter overhead from 600% to 16% compared to fully disentangled counterparts.

---

### In Situ Training of Implicit Neural Compressors for Scientific Simulations via Sketch-Based Regularization

Authors: Cooper Simpson, Stephen Becker, Alireza Doostan
Published: 2025-11-04
Categories: cs.LG, cs.AI, cs.CE, math.NA
arXiv: https://arxiv.org/abs/2511.02659v4

Focusing on implicit neural representations, we present a novel in situ training protocol that employs limited memory buffers of full and sketched data samples, where the sketched data are leveraged to prevent catastrophic forgetting. The theoretical motivation for our use of sketching as a regularizer is presented via a simple Johnson-Lindenstrauss-informed result. While our methods may be of wider interest in the field of continual learning, we specifically target in situ neural compression using implicit neural representation-based hypernetworks. We evaluate our method on a variety of complex simulation data in two and three dimensions, over long time horizons, and across unstructured grids and non-Cartesian geometries. On these tasks, we show strong reconstruction performance at high compression rates. Most importantly, we demonstrate that sketching enables the presented in situ scheme to approximately match the performance of the equivalent offline method.

---

### In Situ Training of Implicit Neural Compressors for Scientific Simulations via Sketch-Based Regularization

Authors: Cooper Simpson, Stephen Becker, Alireza Doostan
Published: 2025-11-04
Categories: cs.LG, cs.AI, cs.CE, cs.NA, math.NA
arXiv: https://arxiv.org/abs/2511.02659v2

Focusing on implicit neural representations, we present a novel in situ training protocol that employs limited memory buffers of full and sketched data samples, where the sketched data are leveraged to prevent catastrophic forgetting. The theoretical motivation for our use of sketching as a regularizer is presented via a simple Johnson-Lindenstrauss-informed result. While our methods may be of wider interest in the field of continual learning, we specifically target in situ neural compression using implicit neural representation-based hypernetworks. We evaluate our method on a variety of complex simulation data in two and three dimensions, over long time horizons, and across unstructured grids and non-Cartesian geometries. On these tasks, we show strong reconstruction performance at high compression rates. Most importantly, we demonstrate that sketching enables the presented in situ scheme to approximately match the performance of the equivalent offline method.

---

### A Byzantine-Robust Federated Learning with Nesterov Momentum

Authors: Lihan Xu, Xiaoyi Fan, Gang Wang, Runhao Zeng, Xiping Hu, Yanjie Dong
Published: 2025-11-04
Categories: cs.LG
arXiv: https://arxiv.org/abs/2511.02657v4

We investigate robust federated learning, where a group of workers collaboratively train a shared model under the orchestration of a central server in the presence of Byzantine adversaries capable of arbitrary and potentially malicious behaviors. To improve communication efficiency while retaining resilience against such adversaries, we propose a Byzantine-resilient Nesterov-augmented federated learning (Byrd-NAFL) algorithm. Byrd-NAFL applies Nesterov's momentum at the server to the output of any soft $γ$-Byzantine resilient aggregation rule, while each worker only computes and uploads a stochastic gradient, which keeps the workers stateless, as desired for intermittently participating Internet of Things devices. For smooth non-convex losses, we establish a finite-time convergence guarantee for all momentum factors $β\in(0,1)$ without requiring the aggregated gradient to be unbiased or bounded; the bound reduces to that of Byzantine-resilient stochastic gradient descent at $β=0$ and is tighter than that of server-side heavy-ball momentum. Experiments on MNIST, CIFAR-10, and Sentiment140 show that Byrd-NAFL converges faster and attains higher accuracy than existing benchmarks in most i.i.d. and non-i.i.d. setups under diverse attacks, while its resilience is bounded by that of the underlying aggregation rule.

---

### Nesterov-Accelerated Robust Federated Learning Over Byzantine Adversaries

Authors: Lihan Xu, Yanjie Dong, Gang Wang, Runhao Zeng, Xiaoyi Fan, Xiping Hu
Published: 2025-11-04
Categories: cs.LG
arXiv: https://arxiv.org/abs/2511.02657v1

We investigate robust federated learning, where a group of workers collaboratively train a shared model under the orchestration of a central server in the presence of Byzantine adversaries capable of arbitrary and potentially malicious behaviors. To simultaneously enhance communication efficiency and robustness against such adversaries, we propose a Byzantine-resilient Nesterov-Accelerated Federated Learning (Byrd-NAFL) algorithm. Byrd-NAFL seamlessly integrates Nesterov's momentum into the federated learning process alongside Byzantine-resilient aggregation rules to achieve fast and safeguarding convergence against gradient corruption. We establish a finite-time convergence guarantee for Byrd-NAFL under non-convex and smooth loss functions with relaxed assumption on the aggregated gradients. Extensive numerical experiments validate the effectiveness of Byrd-NAFL and demonstrate the superiority over existing benchmarks in terms of convergence speed, accuracy, and resilience to diverse Byzantine attack strategies.

---

### Differentiable Hierarchical Visual Tokenization

Authors: Marius Aasan, Martine Hjelkrem-Tan, Nico Catalano, Changkyu Choi, Adín Ramírez Rivera
Published: 2025-11-04
Categories: cs.CV
arXiv: https://arxiv.org/abs/2511.02652v1

Vision Transformers rely on fixed patch tokens that ignore the spatial and semantic structure of images. In this work, we introduce an end-to-end differentiable tokenizer that adapts to image content with pixel-level granularity while remaining backward-compatible with existing architectures for retrofitting pretrained models. Our method uses hierarchical model selection with information criteria to provide competitive performance in both image-level classification and dense-prediction tasks, and even supports out-of-the-box raster-to-vector conversion.

---

### Apriel-H1: Towards Efficient Enterprise Reasoning Models

Authors: Oleksiy Ostapenko, Luke Kumar, Raymond Li, Denis Kocetkov, Joel Lamy-Poirier, Shruthan Radhakrishna, Soham Parikh, Shambhavi Mishra, Sebastien Paquet, Srinivas Sunkara, Valérie Bécaert, Sathwik Tejaswi Madhusudhan et al.
Published: 2025-11-04
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2511.02651v1

Large Language Models (LLMs) achieve remarkable reasoning capabilities through transformer architectures with attention mechanisms. However, transformers suffer from quadratic time and memory complexity in the attention module (MHA) and require caching key-value states during inference, which severely limits throughput and scalability. High inference throughput is critical for agentic tasks, long-context reasoning, efficient deployment under high request loads, and more efficient test-time compute scaling. State Space Models (SSMs) such as Mamba offer a promising alternative with linear inference complexity and a constant memory footprint via recurrent computation with fixed-size hidden states. In this technical report we introduce the Apriel-H1 family of hybrid LLMs that combine transformer attention and SSM sequence mixers for efficient reasoning at 15B model size. These models are obtained through incremental distillation from a pretrained reasoning transformer, Apriel-Nemotron-15B-Thinker, progressively replacing less critical attention layers with linear Mamba blocks. We release multiple post-distillation variants of Apriel-H1-15B-Thinker with different SSM-to-MHA ratios and analyse how reasoning performance degrades as more Mamba layers replace MHA. Additionally, we release a 30/50 hybrid variant of Apriel-H1, further fine-tuned on a supervised dataset of reasoning traces, achieving over 2x higher inference throughput when deployed in the production-ready vLLM environment, with minimal degradation in reasoning performance. This shows that distilled hybrid SSM-Transformer architectures can deliver substantial efficiency gains over the pretrained transformer equivalent without substantially compromising the reasoning quality.

---

### Can Visual Input Be Compressed? A Visual Token Compression Benchmark for Large Multimodal Models

Authors: Tianfan Peng, Yuntao Du, Pengzhou Ji, Shijie Dong, Kailin Jiang, Mingchuan Ma, Yijun Tian, Jinhe Bi, Qian Li, Wei Du, Feng Xiao, Lizhen Cui
Published: 2025-11-04
Categories: cs.CV
arXiv: https://arxiv.org/abs/2511.02650v2

Large multimodal models (LMMs) often suffer from severe inference inefficiency due to the large number of visual tokens introduced by image encoders. While recent token compression methods, such as pruning and merging, have shown promise in reducing redundancy, their evaluation remains fragmented and inconsistent. In this work, we present UniPruneBench, a unified and extensible benchmark for visual token pruning in multimodal LLMs. UniPruneBench provides standardized protocols across six ability dimensions and ten datasets, covering ten representative compression algorithms and three families of LMMs (LLaVA-v1.5, Intern-VL3, and Qwen2.5-VL). Beyond task accuracy, it incorporates system-level metrics such as runtime and prefilling latency to provide a holistic view. Our experiments uncover several key findings: (1) random pruning is a surprisingly strong baseline, (2) no single method consistently outperforms others across scenarios, (3) pruning sensitivity varies significantly across tasks, with OCR being most vulnerable, and (4) pruning ratio is the dominant factor governing performance degradation. We believe UniPruneBench will serve as a reliable foundation for future research on efficient multimodal modeling.

---

### Can Visual Input Be Compressed? A Visual Token Compression Benchmark for Large Multimodal Models

Authors: Tianfan Peng, Yuntao Du, Pengzhou Ji, Shijie Dong, Kailin Jiang, Mingchuan Ma, Yijun Tian, Jinhe Bi, Qian Li, Wei Du, Feng Xiao, Lizhen Cui
Published: 2025-11-04
Categories: cs.CV
arXiv: https://arxiv.org/abs/2511.02650v1

Large multimodal models (LMMs) often suffer from severe inference inefficiency due to the large number of visual tokens introduced by image encoders. While recent token compression methods, such as pruning and merging, have shown promise in reducing redundancy, their evaluation remains fragmented and inconsistent. In this work, we present UniPruneBench, a unified and extensible benchmark for visual token pruning in multimodal LLMs. UniPruneBench provides standardized protocols across six ability dimensions and ten datasets, covering ten representative compression algorithms and three families of LMMs (LLaVA-v1.5, Intern-VL3, and Qwen2.5-VL). Beyond task accuracy, it incorporates system-level metrics such as runtime and prefilling latency to provide a holistic view. Our experiments uncover several key findings: (1) random pruning is a surprisingly strong baseline, (2) no single method consistently outperforms others across scenarios, (3) pruning sensitivity varies significantly across tasks, with OCR being most vulnerable, and (4) pruning ratio is the dominant factor governing performance degradation. We believe UniPruneBench will serve as a reliable foundation for future research on efficient multimodal modeling.

---

### Federated Attention: A Distributed Paradigm for Collaborative LLM Inference over Edge Networks

Authors: Xiumei Deng, Zehui Xiong, Binbin Chen, Dong In Kim, Merouane Debbah, H. Vincent Poor
Published: 2025-11-04
Categories: cs.DC, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2511.02647v1

Large language models (LLMs) are proliferating rapidly at the edge, delivering intelligent capabilities across diverse application scenarios. However, their practical deployment in collaborative scenarios confronts fundamental challenges: privacy vulnerabilities, communication overhead, and computational bottlenecks. To address these, we propose Federated Attention (FedAttn), which integrates the federated paradigm into the self-attention mechanism, creating a new distributed LLM inference framework that simultaneously achieves privacy protection, communication efficiency, and computational efficiency. FedAttn enables participants to perform local self-attention over their own token representations while periodically exchanging and aggregating Key-Value (KV) matrices across multiple Transformer blocks, collaboratively generating LLM responses without exposing private prompts. Further, we identify a structural duality between contextual representation refinement in FedAttn and parameter optimization in FL across private data, local computation, and global aggregation. This key insight provides a principled foundation for systematically porting federated optimization techniques to collaborative LLM inference. Building on this framework, we theoretically analyze how local self-attention computation within participants and heterogeneous token relevance among participants shape error propagation dynamics across Transformer blocks. Moreover, we characterize the fundamental trade-off between response quality and communication/computation efficiency, which is governed by the synchronization interval and the number of participants. Experimental results validate our theoretical analysis, and reveal significant optimization opportunities through sparse attention and adaptive KV aggregation, highlighting FedAttn's potential to deliver scalability and efficiency in real-world edge deployments.

---

### Natural-gas storage modelling by deep reinforcement learning

Authors: Tiziano Balaconi, Aldo Glielmo, Marco Taboga
Published: 2025-11-04
Categories: cs.LG, cs.AI, cs.CE, econ.GN, eess.SY
arXiv: https://arxiv.org/abs/2511.02646v1

We introduce GasRL, a simulator that couples a calibrated representation of the natural gas market with a model of storage-operator policies trained with deep reinforcement learning (RL). We use it to analyse how optimal stockpile management affects equilibrium prices and the dynamics of demand and supply. We test various RL algorithms and find that Soft Actor Critic (SAC) exhibits superior performance in the GasRL environment: multiple objectives of storage operators - including profitability, robust market clearing and price stabilisation - are successfully achieved. Moreover, the equilibrium price dynamics induced by SAC-derived optimal policies have characteristics, such as volatility and seasonality, that closely match those of real-world prices. Remarkably, this adherence to the historical distribution of prices is obtained without explicitly calibrating the model to price data. We show how the simulator can be used to assess the effects of EU-mandated minimum storage thresholds. We find that such thresholds have a positive effect on market resilience against unanticipated shifts in the distribution of supply shocks. For example, with unusually large shocks, market disruptions are averted more often if a threshold is in place.

---

### Robust Face Liveness Detection for Biometric Authentication using Single Image

Authors: Poulami Raha, Yeongnam Chae
Published: 2025-11-04
Categories: cs.CV
arXiv: https://arxiv.org/abs/2511.02645v1

Biometric technologies are widely adopted in security, legal, and financial systems. Face recognition can authenticate a person based on the unique facial features such as shape and texture. However, recent works have demonstrated the vulnerability of Face Recognition Systems (FRS) towards presentation attacks. Using spoofing (aka.,presentation attacks), a malicious actor can get illegitimate access to secure systems. This paper proposes a novel light-weight CNN framework to identify print/display, video and wrap attacks. The proposed robust architecture provides seamless liveness detection ensuring faster biometric authentication (1-2 seconds on CPU). Further, this also presents a newly created 2D spoof attack dataset consisting of more than 500 videos collected from 60 subjects. To validate the effectiveness of this architecture, we provide a demonstration video depicting print/display, video and wrap attack detection approaches. The demo can be viewed in the following link: https://rak.box.com/s/m1uf31fn5amtjp4mkgf1huh4ykfeibaa

---

### Recursively Enumerably Representable Classes and Computable Versions of the Fundamental Theorem of Statistical Learning

Authors: David Kattermann, Lothar Sebastian Krapp
Published: 2025-11-04
Categories: cs.LG, cs.CC, math.LO
arXiv: https://arxiv.org/abs/2511.02644v2

We study computable probably approximately correct (CPAC) learning, where learners are required to be computable functions. It had been previously observed that the Fundamental Theorem of Statistical Learning, which characterizes PAC learnability by finiteness of the Vapnik-Chervonenkis (VC-)dimension, no longer holds in this framework. Recent works recovered analogs of the Fundamental Theorem in the computable setting, for instance by introducing an effective VC-dimension. Guided by this, we investigate the connection between CPAC learning and recursively enumerable representable (RER) classes, whose members can be algorithmically listed. Our results show that the effective VC-dimensions can take arbitrary values above the traditional one, even for RER classes, which creates a whole family of (non-)examples for various notions of CPAC learning. Yet the two dimensions coincide for classes satisfying sufficiently strong notions of CPAC learning. We then observe that CPAC learnability can also be characterized via containment of RER classes that realize the same samples. Furthermore, it is shown that CPAC learnable classes satisfying a unique identification property are necessarily RER. Finally, we establish that agnostic learnability can be guaranteed for RER classes, by considering the relaxed notion of nonuniform CPAC learning.

---

### Recursively Enumerably Representable Classes and Computable Versions of the Fundamental Theorem of Statistical Learning

Authors: David Kattermann, Lothar Sebastian Krapp
Published: 2025-11-04
Categories: cs.LG, cs.CC, math.LO, 8T05, 03D80, 03D25 (Primary) 68Q32, 68T09, 68T27, 68Q04, 03D32
  (Secondary)
arXiv: https://arxiv.org/abs/2511.02644v1

We study computable probably approximately correct (CPAC) learning, where learners are required to be computable functions. It had been previously observed that the Fundamental Theorem of Statistical Learning, which characterizes PAC learnability by finiteness of the Vapnik-Chervonenkis (VC-)dimension, no longer holds in this framework. Recent works recovered analogs of the Fundamental Theorem in the computable setting, for instance by introducing an effective VC-dimension. Guided by this, we investigate the connection between CPAC learning and recursively enumerable representable (RER) classes, whose members can be algorithmically listed. Our results show that the effective VC-dimensions can take arbitrary values above the traditional one, even for RER classes, which creates a whole family of (non-)examples for various notions of CPAC learning. Yet the two dimensions coincide for classes satisfying sufficiently strong notions of CPAC learning. We then observe that CPAC learnability can also be characterized via containment of RER classes that realize the same samples. Furthermore, it is shown that CPAC learnable classes satisfying a unique identification property are necessarily RER. Finally, we establish that agnostic learnability can be guaranteed for RER classes, by considering the relaxed notion of nonuniform CPAC learning.

---

### DecompSR: A dataset for decomposed analyses of compositional multihop spatial reasoning

Authors: Lachlan McPheat, Navdeep Kaur, Robert Blackwell, Alessandra Russo, Anthony G. Cohn, Pranava Madhyastha
Published: 2025-11-04
Categories: cs.AI
arXiv: https://arxiv.org/abs/2511.02627v4

We introduce DecompSR, decomposed spatial reasoning, a large benchmark dataset (over 5m datapoints) and generation framework designed to analyse compositional spatial reasoning ability. The generation of DecompSR allows users to independently vary several aspects of compositionality, namely: productivity (reasoning depth), substitutivity (entity and linguistic variability), overgeneralisation (input order, distractors) and systematicity (novel linguistic elements). DecompSR is built procedurally in a manner which makes it is correct by construction, which is independently verified using a symbolic solver to guarantee the correctness of the dataset. DecompSR is comprehensively benchmarked across a host of Large Language Models (LLMs) where we show that LLMs struggle with productive and systematic generalisation in spatial reasoning tasks whereas they are more robust to linguistic variation. DecompSR provides a provably correct and rigorous benchmarking dataset with a novel ability to independently vary the degrees of several key aspects of compositionality, allowing for robust and fine-grained probing of the compositional reasoning abilities of LLMs.

---

### DecompSR: A dataset for decomposed analyses of compositional multihop spatial reasoning

Authors: Lachlan McPheat, Navdeep Kaur, Robert Blackwell, Alessandra Russo, Anthony G. Cohn, Pranava Madhyastha
Published: 2025-11-04
Categories: cs.AI
arXiv: https://arxiv.org/abs/2511.02627v1

We introduce DecompSR, decomposed spatial reasoning, a large benchmark dataset (over 5m datapoints) and generation framework designed to analyse compositional spatial reasoning ability. The generation of DecompSR allows users to independently vary several aspects of compositionality, namely: productivity (reasoning depth), substitutivity (entity and linguistic variability), overgeneralisation (input order, distractors) and systematicity (novel linguistic elements). DecompSR is built procedurally in a manner which makes it is correct by construction, which is independently verified using a symbolic solver to guarantee the correctness of the dataset. DecompSR is comprehensively benchmarked across a host of Large Language Models (LLMs) where we show that LLMs struggle with productive and systematic generalisation in spatial reasoning tasks whereas they are more robust to linguistic variation. DecompSR provides a provably correct and rigorous benchmarking dataset with a novel ability to independently vary the degrees of several key aspects of compositionality, allowing for robust and fine-grained probing of the compositional reasoning abilities of LLMs.

---

### Understanding New-Knowledge-Induced Factual Hallucinations in LLMs: Analysis and Interpretation

Authors: Renfei Dang, Peng Hu, Zhejian Lai, Changjiang Gao, Min Zhang, Shujian Huang
Published: 2025-11-04
Categories: cs.CL
arXiv: https://arxiv.org/abs/2511.02626v3

Prior works have shown that fine-tuning on new knowledge can induce factual hallucinations in large language models (LLMs), leading to incorrect outputs when evaluated on previously known information. However, the specific manifestations of such hallucination and its underlying mechanisms remain insufficiently understood. Our work addresses this gap by designing a controlled dataset \textit{Biography-Reasoning}, and conducting a fine-grained analysis across multiple knowledge types and two task types, including knowledge question answering (QA) and knowledge reasoning tasks. We find that hallucinations not only severely affect tasks involving newly introduced knowledge, but also propagate to other evaluation tasks. Moreover, when fine-tuning on a dataset in which a specific knowledge type consists entirely of new knowledge, LLMs exhibit elevated hallucination tendencies. This suggests that the degree of unfamiliarity within a particular knowledge type, rather than the overall proportion of new knowledge, is a stronger driver of hallucinations. Through interpretability analysis, we show that learning new knowledge weakens the model's attention to key entities in the input question, leading to an over-reliance on surrounding context and a higher risk of hallucination. Conversely, reintroducing a small amount of known knowledge during the later stages of training restores attention to key entities and substantially mitigates hallucination behavior. Finally, we demonstrate that disrupted attention patterns can propagate across lexically similar contexts, facilitating the spread of hallucinations beyond the original task.

---

### Understanding New-Knowledge-Induced Factual Hallucinations in LLMs: Analysis, Solution, and Interpretation

Authors: Renfei Dang, Peng Hu, Changjiang Gao, Shujian Huang
Published: 2025-11-04
Categories: cs.CL
arXiv: https://arxiv.org/abs/2511.02626v1

Previous studies show that introducing new knowledge during large language models (LLMs) fine-tuning can lead to the generation of erroneous output when tested on known information, thereby triggering factual hallucinations. However, existing studies have not deeply investigated the specific manifestations and underlying mechanisms of these hallucinations. Our work addresses this gap by designing a controlled dataset Biography-Reasoning, and conducting a fine-grained analysis across multiple knowledge types and two task types, including knowledge question answering (QA) and knowledge reasoning tasks. We find that when fine-tuned on a dataset in which a specific knowledge type consists entirely of new knowledge, LLMs exhibit significantly increased hallucination tendencies. This suggests that the high unfamiliarity of a particular knowledge type, rather than the overall proportion of new knowledge, is a stronger driver of hallucinations, and these tendencies can even affect other knowledge types in QA tasks. To mitigate such factual hallucinations, we propose KnownPatch, which patches a small number of known knowledge samples in the later stages of training, effectively alleviating new-knowledge-induced hallucinations. Through attention analysis, we find that learning new knowledge reduces the model's attention to key entities in the question, thus causing excessive focus on the surrounding context, which may increase the risk of hallucination. Moreover, the attention pattern can propagate to similar contexts, facilitating the spread of hallucinations to textually similar questions. Our method effectively mitigates the disruption of new knowledge learning to the model's attention on key entities, accompanied by improved performance.

---

### Condition Numbers and Eigenvalue Spectra of Shallow Networks on Spheres

Authors: Xinliang Liu, Tong Mao, Jinchao Xu
Published: 2025-11-04
Categories: math.NA, cs.LG
arXiv: https://arxiv.org/abs/2511.02625v2

We present an estimation of the condition numbers of the \emph{mass} and \emph{stiffness} matrices arising from shallow ReLU$^k$ neural networks defined on the unit sphere~$\mathbb{S}^d$. In particular, when $\{θ_j^*\}_{j=1}^n \subset \mathbb{S}^d$ is \emph{antipodally quasi-uniform}, the condition number is sharp. Indeed, in this case, we obtain sharp asymptotic estimates for the full spectrum of eigenvalues and characterize the structure of the corresponding eigenspaces, showing that the smallest eigenvalues are associated with an eigenbasis of low-degree polynomials while the largest eigenvalues are linked to high-degree polynomials. This spectral analysis establishes a precise correspondence between the approximation power of the network and its numerical stability.

---

### The stability of shallow neural networks on spheres: A sharp spectral analysis

Authors: Xinliang Liu, Tong Mao, Jinchao Xu
Published: 2025-11-04
Categories: math.NA, cs.LG, cs.NA
arXiv: https://arxiv.org/abs/2511.02625v1

We present an estimation of the condition numbers of the \emph{mass} and \emph{stiffness} matrices arising from shallow ReLU$^k$ neural networks defined on the unit sphere~$\mathbb{S}^d$. In particular, when $\{\theta_j^*\}_{j=1}^n \subset \mathbb{S}^d$ is \emph{antipodally quasi-uniform}, the condition number is sharp. Indeed, in this case, we obtain sharp asymptotic estimates for the full spectrum of eigenvalues and characterize the structure of the corresponding eigenspaces, showing that the smallest eigenvalues are associated with an eigenbasis of low-degree polynomials while the largest eigenvalues are linked to high-degree polynomials. This spectral analysis establishes a precise correspondence between the approximation power of the network and its numerical stability.

---

### The Realignment Problem: When Right becomes Wrong in LLMs

Authors: Aakash Sen Sharma, Debdeep Sanyal, Manodeep Ray, Vivek Srivastava, Shirish Karande, Murari Mandal
Published: 2025-11-04
Categories: cs.CL
arXiv: https://arxiv.org/abs/2511.02623v2

Post-training alignment of large language models (LLMs) relies on large-scale human annotations guided by policy specifications that change over time. Cultural shifts, value reinterpretations, and regulatory or industrial updates make static alignment increasingly brittle. As policies evolve, deployed models can diverge from current alignment objectives, creating an Alignment-Reality Gap that is difficult to audit or correct. Existing remediation typically requires re-annotation under revised guidelines, which introduces systematic challenges, including guideline ambiguity, annotator interpretation drift, and reduced consistency at scale. We introduce TRACE (Triage and Re-align by Alignment Conflict Evaluation), a framework that transforms realignment into a structured optimization problem over existing data without requiring fresh human annotation. Leveraging a stronger model as a proxy judge, TRACE operates via a three-stage pipeline: (1) triaging preference pairs into inversion, suppression, or retention categories based on alignment conflicts; (2) computing an alignment impact score via bi-level optimization to prioritize high-leverage samples; and (3) executing updates using a hybrid objective that combines relational losses (e.g., IPO) for preference inversion and punitive losses (e.g., NPO) for response suppression. Experiments on Qwen2.5-7B, Gemma-2-9B, and Llama-3.1-8B demonstrate robust realignment on synthetic benchmarks and the PKU-SafeRLHF dataset without degrading general utility. This work provides a scalable approach for LLM realignment under evolving data annotation policies and alignment guidelines. We release our code: https://respailab.github.io/TRACE/

---

### The Realignment Problem: When Right becomes Wrong in LLMs

Authors: Aakash Sen Sharma, Debdeep Sanyal, Vivek Srivastava, Shirish Karande, Murari Mandal
Published: 2025-11-04
Categories: cs.CL
arXiv: https://arxiv.org/abs/2511.02623v1

The alignment of Large Language Models (LLMs) with human values is central to their safe deployment, yet current practice produces static, brittle, and costly-to-maintain models that fail to keep pace with evolving norms and policies. This misalignment, which we term the Alignment-Reality Gap, poses a growing challenge for reliable long-term use. Existing remedies are inadequate: large-scale re-annotation is economically prohibitive, and standard unlearning methods act as blunt instruments that erode utility rather than enable precise policy updates. We introduce TRACE (Triage and Re-align by Alignment Conflict Evaluation), a framework for principled unlearning that reconceives re-alignment as a programmatic policy application problem. TRACE programmatically triages existing preference data against a new policy, identifies high-impact conflicts via a alignment impact score, and applies a hybrid optimization that cleanly inverts, discards, or preserves preferences while safeguarding model performance. Empirical results show that TRACE achieves robust re-alignment across diverse model families (Qwen2.5-7B, Gemma-2-9B, Llama-3.1-8B). On both synthetic benchmarks and the PKU-SafeRLHF dataset under complex policy shift, TRACE enforces new principles without degrading general capabilities. Our work establishes a scalable, dynamic, and cost-effective paradigm for maintaining LLM alignment, providing a foundation for sustainable and responsible AI deployment.

---

### Verifying LLM Inference to Detect Model Weight Exfiltration

Authors: Roy Rinberg, Adam Karvonen, Alexander Hoover, Daniel Reuter, Keri Warr
Published: 2025-11-04
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2511.02620v3

As large AI models become increasingly valuable assets, the risk of model weight exfiltration from inference servers grows accordingly. An attacker controlling an inference server may exfiltrate model weights by hiding them within ordinary model responses, a strategy known as steganography. This work investigates how to verify LLM model inference to defend against such attacks and, more broadly, to detect anomalous or buggy behavior during inference. We formalize model weight exfiltration as a security game, propose a verification framework that can provably mitigate steganographic exfiltration, and specify the trust assumptions associated with our scheme. To enable verification, we characterize valid sources of non-determinism in large language model inference and introduce two practical estimators for them. We evaluate our detection framework on several open-weight models ranging from 3B to 30B parameters. On MOE-Qwen-30B, our detector reduces exfiltratable information to <0.5% with false-positive rate of <0.01%, corresponding to a >200x slowdown for adversaries. Overall, this work further establishes a foundation for defending against model weight exfiltration and demonstrates that strong protection can be achieved with minimal additional cost to inference providers. Our code is made public at: https://github.com/RoyRin/inference_verification_for_model_weight_exfiltration .

---

### Verifying LLM Inference to Prevent Model Weight Exfiltration

Authors: Roy Rinberg, Adam Karvonen, Alex Hoover, Daniel Reuter, Keri Warr
Published: 2025-11-04
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2511.02620v1

As large AI models become increasingly valuable assets, the risk of model weight exfiltration from inference servers grows accordingly. An attacker controlling an inference server may exfiltrate model weights by hiding them within ordinary model outputs, a strategy known as steganography. This work investigates how to verify model responses to defend against such attacks and, more broadly, to detect anomalous or buggy behavior during inference. We formalize model exfiltration as a security game, propose a verification framework that can provably mitigate steganographic exfiltration, and specify the trust assumptions associated with our scheme. To enable verification, we characterize valid sources of non-determinism in large language model inference and introduce two practical estimators for them. We evaluate our detection framework on several open-weight models ranging from 3B to 30B parameters. On MOE-Qwen-30B, our detector reduces exfiltratable information to <0.5% with false-positive rate of 0.01%, corresponding to a >200x slowdown for adversaries. Overall, this work further establishes a foundation for defending against model weight exfiltration and demonstrates that strong protection can be achieved with minimal additional cost to inference providers.

---

### A Non-Adversarial Approach to Idempotent Generative Modelling

Authors: Mohammed Al-Jaff, Giovanni Luca Marchetti, Michael C Welle, Jens Lundell, Mats G. Gustafsson, Gustav Eje Henter, Hossein Azizpour, Danica Kragic
Published: 2025-11-04
Categories: cs.LG
arXiv: https://arxiv.org/abs/2511.02614v1

Idempotent Generative Networks (IGNs) are deep generative models that also function as local data manifold projectors, mapping arbitrary inputs back onto the manifold. They are trained to act as identity operators on the data and as idempotent operators off the data manifold. However, IGNs suffer from mode collapse, mode dropping, and training instability due to their objectives, which contain adversarial components and can cause the model to cover the data manifold only partially -- an issue shared with generative adversarial networks. We introduce Non-Adversarial Idempotent Generative Networks (NAIGNs) to address these issues. Our loss function combines reconstruction with the non-adversarial generative objective of Implicit Maximum Likelihood Estimation (IMLE). This improves on IGN's ability to restore corrupted data and generate new samples that closely match the data distribution. We moreover demonstrate that NAIGNs implicitly learn the distance field to the data manifold, as well as an energy-based model.

---

### Towards Migrating Neural Network Implementations

Authors: Nadia Daoudi, Ivan Alfonso, Jordi Cabot
Published: 2025-11-04
Categories: cs.LG
arXiv: https://arxiv.org/abs/2511.02610v2

The development of smart systems (i.e., systems enhanced with AI components) has thrived thanks to the rapid advancements in neural networks (NNs). A wide range of libraries and frameworks have consequently emerged to support NN design and implementation. The choice depends on factors such as available functionalities, ease of use, documentation and community support. After adopting a given NN framework, organizations might later choose to switch to another if performance declines, requirements evolve, or new features are introduced. Unfortunately, migrating NN implementations across libraries is challenging due to the lack of migration approaches specifically tailored for NNs. This leads to increased time and effort to modernize NNs, as manual updates are necessary to avoid relying on outdated implementations and ensure compatibility with new features. In this paper, we propose an approach to automatically migrate neural network code across deep learning frameworks. Our method makes use of a pivot NN model to create an abstraction of the NN prior to migration. We validate our approach using two popular NN frameworks, namely PyTorch and TensorFlow. We also discuss the challenges of migrating code between the two frameworks and how they were approached in our method. Experimental evaluation on five NNs shows that our approach successfully migrates their code and produces NNs that are functionally equivalent to the originals. Artefacts from our work are available online.

---

### Neural Network Interoperability Across Platforms

Authors: Nadia Daoudi, Ivan Alfonso, Jordi Cabot
Published: 2025-11-04
Categories: cs.LG
arXiv: https://arxiv.org/abs/2511.02610v1

The development of smart systems (i.e., systems enhanced with AI components) has thrived thanks to the rapid advancements in neural networks (NNs). A wide range of libraries and frameworks have consequently emerged to support NN design and implementation. The choice depends on factors such as available functionalities, ease of use, documentation and community support. After adopting a given NN framework, organizations might later choose to switch to another if performance declines, requirements evolve, or new features are introduced. Unfortunately, migrating NN implementations across libraries is challenging due to the lack of migration approaches specifically tailored for NNs. This leads to increased time and effort to modernize NNs, as manual updates are necessary to avoid relying on outdated implementations and ensure compatibility with new features. In this paper, we propose an approach to automatically migrate neural network code across deep learning frameworks. Our method makes use of a pivot NN model to create an abstraction of the NN prior to migration. We validate our approach using two popular NN frameworks, namely PyTorch and TensorFlow. We also discuss the challenges of migrating code between the two frameworks and how they were approached in our method. Experimental evaluation on five NNs shows that our approach successfully migrates their code and produces NNs that are functionally equivalent to the originals. Artefacts from our work are available online.

---