# Research Papers — 2025-12 · bundle page 137 of 175

> Full-text markdown of papers #6801–6850 of 8703 published in 2025-12, newest first. AI-readable mirrors of the full 8703-paper 2025-12 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2025-12/page/137
[← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/136.md)
[Next bundle →](https://aipod.dev/papers/archive/2025-12/page/138.md)

### The Forgotten Shield: Safety Grafting in Parameter-Space for Medical MLLMs

Authors: Jiale Zhao, Xing Mou, Jinlin Wu, Hongyuan Yu, Mingrui Sun, Yang Shi, Xuanwu Yin, Zhen Chen, Zhen Lei, Yaohua Wang
Published: 2025-12-05
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2601.04199v1

Medical Multimodal Large Language Models (Medical MLLMs) have achieved remarkable progress in specialized medical tasks; however, research into their safety has lagged, posing potential risks for real-world deployment. In this paper, we first establish a multidimensional evaluation framework to systematically benchmark the safety of current SOTA Medical MLLMs. Our empirical analysis reveals pervasive vulnerabilities across both general and medical-specific safety dimensions in existing models, particularly highlighting their fragility against cross-modality jailbreak attacks. Furthermore, we find that the medical fine-tuning process frequently induces catastrophic forgetting of the model's original safety alignment. To address this challenge, we propose a novel "Parameter-Space Intervention" approach for efficient safety re-alignment. This method extracts intrinsic safety knowledge representations from original base models and concurrently injects them into the target model during the construction of medical capabilities. Additionally, we design a fine-grained parameter search algorithm to achieve an optimal trade-off between safety and medical performance. Experimental results demonstrate that our approach significantly bolsters the safety guardrails of Medical MLLMs without relying on additional domain-specific safety data, while minimizing degradation to core medical performance.

---

### ReCollab: Retrieval-Augmented LLMs for Cooperative Ad-hoc Teammate Modeling

Authors: Conor Wallace, Umer Siddique, Yongcan Cao
Published: 2025-12-05
Categories: cs.MA, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2512.22129v1

Ad-hoc teamwork (AHT) requires agents to infer the behavior of previously unseen teammates and adapt their policy accordingly. Conventional approaches often rely on fixed probabilistic models or classifiers, which can be brittle under partial observability and limited interaction. Large language models (LLMs) offer a flexible alternative: by mapping short behavioral traces into high-level hypotheses, they can serve as world models over teammate behavior. We introduce \Collab, a language-based framework that classifies partner types using a behavior rubric derived from trajectory features, and extend it to \ReCollab, which incorporates retrieval-augmented generation (RAG) to stabilize inference with exemplar trajectories. In the cooperative Overcooked environment, \Collab effectively distinguishes teammate types, while \ReCollab consistently improves adaptation across layouts, achieving Pareto-optimal trade-offs between classification accuracy and episodic return. These findings demonstrate the potential of LLMs as behavioral world models for AHT and highlight the importance of retrieval grounding in challenging coordination settings.

---

### Deep Reinforcement Learning Optimization for Uncertain Nonlinear Systems via Event-Triggered Robust Adaptive Dynamic Programming

Authors: Ningwei Bai, Chi Pui Chan, Qichen Yin, Tengyang Gong, Yunda Yan, Zezhi Tang
Published: 2025-12-05
Categories: math.OC, cs.AI, eess.SY
arXiv: https://arxiv.org/abs/2512.15735v4

This work proposes a unified control architecture that couples a Reinforcement Learning (RL)-driven controller with a disturbance-rejection Extended State Observer (ESO), complemented by an Event-Triggered Mechanism (ETM) to limit unnecessary computations. The ESO is utilized to estimate the system states and the lumped disturbance in real time, forming the foundation for effective disturbance compensation. To obtain near-optimal behavior without an accurate system description, a value-iteration-based Adaptive Dynamic Programming (ADP) method is adopted for policy approximation. The inclusion of the ETM ensures that parameter updates of the learning module are executed only when the state deviation surpasses a predefined bound, thereby preventing excessive learning activity and substantially reducing computational load. A Lyapunov-oriented analysis is used to characterize the stability properties of the resulting closed-loop system. Numerical experiments further confirm that the developed approach maintains strong control performance and disturbance tolerance, while achieving a significant reduction in sampling and processing effort compared with standard time-triggered ADP schemes.

---

### A Context-Free Smart Grid Model Using Complex System Approach

Authors: Soufian Ben Amor, Alain Bui, Guillaume Guerard
Published: 2025-12-05
Categories: math.OC, cs.AI, eess.SY
arXiv: https://arxiv.org/abs/2512.15733v1

Energy and pollution are urging problems of the 21th century. By gradually changing the actual power grid system, smart grid may evolve into different systems by means of size, elements and strategies, but its fundamental requirements and objectives will not change such as optimizing production, transmission, and consumption. Studying the smart grid through modeling and simulation provides us with valuable results which cannot be obtained in real world due to time and cost related constraints. Moreover, due to the complexity of the smart grid, achieving global optimization is not an easy task. In this paper, we propose a complex system based approach to the smart grid modeling, accentuating on the optimization by combining game theoretical and classical methods in different levels. Thanks to this combination, the optimization can be achieved with flexibility and scalability, while keeping its generality.

---

### The Red Queen's Trap: Limits of Deep Evolution in High-Frequency Trading

Authors: Yijia Chen
Published: 2025-12-05
Categories: q-fin.TR, cs.LG, cs.NE, q-fin.CP
arXiv: https://arxiv.org/abs/2512.15732v1

The integration of Deep Reinforcement Learning (DRL) and Evolutionary Computation (EC) is frequently hypothesized to be the "Holy Grail" of algorithmic trading, promising systems that adapt autonomously to non-stationary market regimes. This paper presents a rigorous post-mortem analysis of "Galaxy Empire," a hybrid framework coupling LSTM/Transformer-based perception with a genetic "Time-is-Life" survival mechanism. Deploying a population of 500 autonomous agents in a high-frequency cryptocurrency environment, we observed a catastrophic divergence between training metrics (Validation APY $>300\%$) and live performance (Capital Decay $>70\%$). We deconstruct this failure through a multi-disciplinary lens, identifying three critical failure modes: the overfitting of \textit{Aleatoric Uncertainty} in low-entropy time-series, the \textit{Survivor Bias} inherent in evolutionary selection under high variance, and the mathematical impossibility of overcoming microstructure friction without order-flow data. Our findings provide empirical evidence that increasing model complexity in the absence of information asymmetry exacerbates systemic fragility.

---

### TinyMyo: a Tiny Foundation Model for Flexible EMG Signal Processing at the Edge

Authors: Matteo Fasulo, Giusy Spacone, Thorir Mar Ingolfsson, Yawei Li, Luca Benini, Andrea Cossettini
Published: 2025-12-05
Categories: eess.SP, cs.AI, cs.HC, cs.LG
arXiv: https://arxiv.org/abs/2512.15729v2

Objective: Surface electromyography (EMG) is a non-invasive sensing modality widely used in biomechanics, rehabilitation, prosthetic control, and human-machine interfaces. Despite decades of use, achieving robust generalization across subjects, recording systems, and acquisition protocols remains challenging. While foundation models (FMs) are gaining traction for EMG, existing approaches remain limited to single downstream tasks and lack deployability on embedded platforms. This work addresses these limitations. Methods: We present TinyMyo, a lightweight FM based on a Transformer encoder architecture. The model is pre-trained in a self-supervised manner using masked reconstruction on publicly available datasets. With only 3.6M parameters, TinyMyo is designed to support multiple downstream tasks through minimal task-specific head adaptations. Results: We demonstrate generalization across hand gesture classification, hand kinematic regression, speech production and speech recognition, with performance comparable to or surpassing the state of the art (SoA), and model size below 5M parameters. We achieve SoA results compared to previous FM-based works on the NinaPro DB5 (89.4%), UCI-EMG (97.56%), and EPN-612 (96.74%) datasets. We demonstrate the first-time deployment of an EMG FM on an ultra-low power microcontroller (GAP9), with an inference time of 0.785 s, energy of 44.91 mJ and power envelope of 57.18 mW. Conclusion: TinyMyo demonstrates that compact, self-supervised EMG FM can guarantee strong generalization across multiple downstream tasks while remaining compatible with low-power edge devices. Significance: TinyMyo is the first EMG FM for ultra-low power edge devices, enabling scalable and energy-efficient sensing for motor intent decoding, neuromuscular assessment, and biosignal driven human-machine interaction.

---

### FedSight AI: Multi-Agent System Architecture for Federal Funds Target Rate Prediction

Authors: Yuhan Hou, Tianji Rao, Jeremy Tan, Adler Viton, Xiyue Zhang, David Ye, Abhishek Kodi, Sanjana Dulam, Aditya Paul, Yikai Feng
Published: 2025-12-05
Categories: q-fin.GN, cs.AI
arXiv: https://arxiv.org/abs/2512.15728v1

The Federal Open Market Committee (FOMC) sets the federal funds rate, shaping monetary policy and the broader economy. We introduce \emph{FedSight AI}, a multi-agent framework that uses large language models (LLMs) to simulate FOMC deliberations and predict policy outcomes. Member agents analyze structured indicators and unstructured inputs such as the Beige Book, debate options, and vote, replicating committee reasoning. A Chain-of-Draft (CoD) extension further improves efficiency and accuracy by enforcing concise multistage reasoning. Evaluated at 2023-2024 meetings, FedSight CoD achieved accuracy of 93.75\% and stability of 93.33\%, outperforming baselines including MiniFed and Ordinal Random Forest (RF), while offering transparent reasoning aligned with real FOMC communications.

---

### Tourists Profiling by Interest Analysis

Authors: Sonia Djebali, Quentin Gabot, Guillaume Guerard
Published: 2025-12-05
Categories: cs.SI, cs.AI
arXiv: https://arxiv.org/abs/2512.14704v1

With the recent digital revolution, analyzing of tourists' behaviors and research fields associated with it have changed profoundly. It is now easier to examine behaviors of tourists using digital traces they leave during their travels. The studies conducted on diverse aspects of tourism focus on quantitative aspects of digital traces to reach its conclusions. In this paper, we suggest a study focused on both qualitative and quantitative aspect of digital traces to understand the dynamics governing tourist behavior, especially those concerning attractions networks.

---

### Modular connectivity in neural networks emerges from Poisson noise-motivated regularisation, and promotes robustness and compositional generalisation

Authors: Daoyuan Qian, Qiyao Liang, Ila Fiete
Published: 2025-12-05
Categories: physics.bio-ph, cs.LG, cs.NE, stat.ML
arXiv: https://arxiv.org/abs/2512.13707v1

Circuits in the brain commonly exhibit modular architectures that factorise complex tasks, resulting in the ability to compositionally generalise and reduce catastrophic forgetting. In contrast, artificial neural networks (ANNs) appear to mix all processing, because modular solutions are difficult to find as they are vanishing subspaces in the space of possible solutions. Here, we draw inspiration from fault-tolerant computation and the Poisson-like firing of real neurons to show that activity-dependent neural noise, combined with nonlinear neural responses, drives the emergence of solutions that reflect an accurate understanding of modular tasks, corresponding to acquisition of a correct world model. We find that noise-driven modularisation can be recapitulated by a deterministic regulariser that multiplicatively combines weights and activations, revealing rich phenomenology not captured in linear networks or by standard regularisation methods. Though the emergence of modular structure requires sufficiently many training samples (exponential in the number of modular task dimensions), we show that pre-modularised ANNs exhibit superior noise-robustness and the ability to generalise and extrapolate well beyond training data, compared to ANNs without such inductive biases. Together, our work demonstrates a regulariser and architectures that could encourage modularity emergence to yield functional benefits.

---

### Mitigating Catastrophic Forgetting in Mathematical Reasoning Finetuning through Mixed Training

Authors: John Graham Reynolds
Published: 2025-12-05
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2512.13706v1

When finetuning large language models for specialized tasks such as mathematical reasoning, models exhibit catastrophic forgetting, losing previously learned capabilities. We investigate this by finetuning Flan-T5-Base (250M parameters) on the DeepMind Mathematics dataset and measuring forgetting on MultiNLI. Math-only training improves mathematical accuracy from 3.1\% to 12.0\% but causes NLI accuracy to collapse from 81.0\% to 16.5\%--a 64.5 percentage point drop occurring within the first 1,000 training steps. We propose mixed training strategies that interleave mathematical and NLI examples during training. Our results demonstrate that mixed training completely eliminates catastrophic forgetting while maintaining equivalent mathematical performance: the balanced 1:1 ratio achieves 12.0\% math accuracy (matching math-only) while preserving 86.2\% NLI accuracy. We systematically explore mixing ratios from 1:1 to 15:1, finding that even minimal NLI exposure (6.2\%) provides effective regularization. These findings demonstrate that specialization need not require forgetting general capabilities, with implications for scaling to larger models where mixed training may confer additional benefits beyond forgetting prevention.

---

### Scaling and Transferability of Annealing Strategies in Large Language Model Training

Authors: Siqi Wang, Zhengyu Chen, Teng Xiao, Zheqi Lv, Jinluan Yang, Xunliang Cai, Jingang Wang, Xiaomeng Li
Published: 2025-12-05
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2512.13705v1

Learning rate scheduling is crucial for training large language models, yet understanding the optimal annealing strategies across different model configurations remains challenging. In this work, we investigate the transferability of annealing dynamics in large language model training and refine a generalized predictive framework for optimizing annealing strategies under the Warmup-Steady-Decay (WSD) scheduler. Our improved framework incorporates training steps, maximum learning rate, and annealing behavior, enabling more efficient optimization of learning rate schedules. Our work provides a practical guidance for selecting optimal annealing strategies without exhaustive hyperparameter searches, demonstrating that smaller models can serve as reliable proxies for optimizing the training dynamics of larger models. We validate our findings on extensive experiments using both Dense and Mixture-of-Experts (MoE) models, demonstrating that optimal annealing ratios follow consistent patterns and can be transferred across different training configurations.

---

### Adjudicator: Correcting Noisy Labels with a KG-Informed Council of LLM Agents

Authors: Doohee You, Sundeep Paul
Published: 2025-12-05
Categories: cs.AI
arXiv: https://arxiv.org/abs/2512.13704v1

The performance of production machine learning systems is fundamentally limited by the quality of their training data. In high-stakes industrial applications, noisy labels can degrade performance and erode user trust. This paper presents Adjudicator, a system that addresses the critical data mining challenge of automatically identifying and correcting label noise and has been validated for production deployment. Adjudicator models this as a neuro-symbolic task, first constructing a dynamic Knowledge Graph (KG) to unify item context. This KG then informs a "Council of Agents," a novel multi-agent Large Language Model architecture where specialized agents debate and vote on a label's validity. We validate our system on a 1,000-item balanced subset of the AlleNoise benchmark. Our KG-informed model achieves a 0.99 F1-score, significantly outperforming a single-LLM baseline (0.48 F1) and a non-KG council (0.59 F1). Our analysis reveals this is due to a Precision, achieved by a novel override logic that uses the KG to perfectly identify complex, structural errors (complete Recall) -- a class of errors that baselines fail to find. This result demonstrates a robust and explainable system for automated, high-precision data verification, serving as a vital proof-of-concept for generating golden datasets in strictly governed industrial environments.

---

### Safe2Harm: Semantic Isomorphism Attacks for Jailbreaking Large Language Models

Authors: Fan Yang
Published: 2025-12-05
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2512.13703v1

Large Language Models (LLMs) have demonstrated exceptional performance across various tasks, but their security vulnerabilities can be exploited by attackers to generate harmful content, causing adverse impacts across various societal domains. Most existing jailbreak methods revolve around Prompt Engineering or adversarial optimization, yet we identify a previously overlooked phenomenon: many harmful scenarios are highly consistent with legitimate ones in terms of underlying principles. Based on this finding, this paper proposes the Safe2Harm Semantic Isomorphism Attack method, which achieves efficient jailbreaking through four stages: first, rewrite the harmful question into a semantically safe question with similar underlying principles; second, extract the thematic mapping relationship between the two; third, let the LLM generate a detailed response targeting the safe question; finally, reversely rewrite the safe response based on the thematic mapping relationship to obtain harmful output. Experiments on 7 mainstream LLMs and three types of benchmark datasets show that Safe2Harm exhibits strong jailbreaking capability, and its overall performance is superior to existing methods. Additionally, we construct a challenging harmful content evaluation dataset containing 358 samples and evaluate the effectiveness of existing harmful detection methods, which can be deployed for LLM input-output filtering to enable defense.

---

### Using Socio-economic Indicators, Smart Transit Systems, and Urban Simulator to Accelerate ZEV Adoption and Reduce VMT

Authors: Mulham Fawakherji, Bruce Race, Driss Benhaddou
Published: 2025-12-05
Categories: cs.CY, cs.AI
arXiv: https://arxiv.org/abs/2512.11870v2

Globally, on-road transportation accounts for 15% of greenhouse gas (GHG) emissions and an estimated 385,000 premature deaths from PM2.5. Cities play a critical role in meeting IPCC targets, generating 75% of global energy-related GHG emissions. In Houston, Texas, on-road transportation represents 48% of baseline emissions in the Climate Action Plan (CAP). To reach net-zero by 2050, the CAP targets a 70% emissions reduction from a 2014 baseline, offset by 30% renewable energy. This goal is challenging because Houston is low-density and auto-dependent, with 89% of on-road emissions from cars and small trucks and limited public transit usage. Socio-economic disparities further constrain Zero Emissions Vehicle (ZEV) adoption. Strategies focus on expanding ZEV access and reducing Vehicle Miles Traveled (VMT) by 20% through transit improvements and city design. This paper presents methods for establishing an on-road emissions baseline and evaluating policies that leverage socio-economic indicators and Intelligent Transportation Systems (ITS) to accelerate ZEV adoption and reduce VMT. Smart parking, transit incentives, secure data systems, and ZEV fleet management support improvements in modal split and system reliability. Policy options are analyzed and potential actions identified. To support evaluation, a simulation environment was developed in Unity 3D, enabling dynamic modeling of urban mobility and visualization of policy scenarios. Auto-dependent cities aiming for 2050 emission targets can benefit from the indicators, metrics, and technologies discussed.

---

### Temporal-Anchor3DLane: Enhanced 3D Lane Detection with Multi-Task Losses and LSTM Fusion

Authors: D. Shainu Suhas, G. Rahul, K. Muni
Published: 2025-12-05
Categories: cs.CV
arXiv: https://arxiv.org/abs/2512.11869v1

Monocular 3D lane detection remains challenging due to depth ambiguity, occlusion, and temporal instability across frames. Anchor-based approaches such as Anchor3DLane have demonstrated strong performance by regressing continuous 3D lane curves from multi-camera surround views. However, the baseline model still exhibits (i) sensitivity to regression outliers, (ii) weak supervision of global curve geometry, (iii) difficulty in balancing multiple loss terms, and (iv) limited exploitation of temporal continuity. We propose Temporal-Anchor3DLane, an enhanced 3D lane detection framework that extends Anchor3DLane with three key contributions: (1) a set of multi-task loss improvements, including Balanced L1 regression, Chamfer point-set distance, and uncertainty-based loss weighting, together with focal and Dice components for classification and visibility; (2) a lightweight Temporal LSTM Fusion module that aggregates per-anchor features across frames, replacing a heavier Transformer-style temporal fusion; and (3) ESCOP-style training refinements that couple curve-level supervision with temporal consistency. On OpenLane, Temporal-Anchor3DLane improves F1 by +6.2 and yields smoother temporal trajectories, showing that small architectural and loss refinements significantly enhance 3D lane robustness without extra sensors or scaling.

---

### Industrial AI Robustness Card for Time Series Models

Authors: Alexander Windmann, Benedikt Stratmann, Mariya Lyashenko, Oliver Niggemann
Published: 2025-12-05
Categories: cs.CY, cs.AI
arXiv: https://arxiv.org/abs/2512.11868v2

Industrial AI practitioners face vague robustness requirements in emerging regulations and standards but lack concrete, implementation-ready protocols. This paper introduces the Industrial AI Robustness Card for Time Series (IARC-TS), a lightweight protocol for documenting and evaluating industrial time series models. IARC-TS specifies required fields and an empirical measurement and reporting protocol that combines drift and operational domain monitoring, uncertainty quantification, and stress tests, and maps these to selected EU AI Act documentation, testing, and monitoring obligations. A biopharmaceutical soft sensor case study illustrates how IARC-TS supports reproducible robustness evidence and defines monitoring triggers.

---

### On the Dangers of Bootstrapping Generation for Continual Learning and Beyond

Authors: Daniil Zverev, A. Sophia Koepke, Joao F. Henriques
Published: 2025-12-05
Categories: cs.LG, cs.AI, cs.CV, eess.IV
arXiv: https://arxiv.org/abs/2512.11867v1

The use of synthetically generated data for training models is becoming a common practice. While generated data can augment the training data, repeated training on synthetic data raises concerns about distribution drift and degradation of performance due to contamination of the dataset. We investigate the consequences of this bootstrapping process through the lens of continual learning, drawing a connection to Generative Experience Replay (GER) methods. We present a statistical analysis showing that synthetic data introduces significant bias and variance into training objectives, weakening the reliability of maximum likelihood estimation. We provide empirical evidence showing that popular generative models collapse under repeated training with synthetic data. We quantify this degradation and show that state-of-the-art GER methods fail to maintain alignment in the latent space. Our findings raise critical concerns about the use of synthetic data in continual learning.

---

### Phase transitions reveal hierarchical structure in deep neural networks

Authors: Ibrahim Talha Ersoy, Andrés Fernando Cardozo Licha, Karoline Wiesner
Published: 2025-12-05
Categories: cs.LG, physics.data-an
arXiv: https://arxiv.org/abs/2512.11866v1

Training Deep Neural Networks relies on the model converging on a high-dimensional, non-convex loss landscape toward a good minimum. Yet, much of the phenomenology of training remains ill understood. We focus on three seemingly disparate observations: the occurrence of phase transitions reminiscent of statistical physics, the ubiquity of saddle points, and phenomenon of mode connectivity relevant for model merging. We unify these within a single explanatory framework, the geometry of the loss and error landscapes. We analytically show that phase transitions in DNN learning are governed by saddle points in the loss landscape. Building on this insight, we introduce a simple, fast, and easy to implement algorithm that uses the L2 regularizer as a tool to probe the geometry of error landscapes. We apply it to confirm mode connectivity in DNNs trained on the MNIST dataset by efficiently finding paths that connect global minima. We then show numerically that saddle points induce transitions between models that encode distinct digit classes. Our work establishes the geometric origin of key training phenomena in DNNs and reveals a hierarchy of accuracy basins analogous to phases in statistical physics.

---

### Explainable Adversarial-Robust Vision-Language-Action Model for Robotic Manipulation

Authors: Ju-Young Kim, Ji-Hong Park, Myeongjun Kim, Gun-Woo Kim
Published: 2025-12-05
Categories: cs.CV, cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2512.11865v1

Smart farming has emerged as a key technology for advancing modern agriculture through automation and intelligent control. However, systems relying on RGB cameras for perception and robotic manipulators for control, common in smart farming, are vulnerable to photometric perturbations such as hue, illumination, and noise changes, which can cause malfunction under adversarial attacks. To address this issue, we propose an explainable adversarial-robust Vision-Language-Action model based on the OpenVLA-OFT framework. The model integrates an Evidence-3 module that detects photometric perturbations and generates natural language explanations of their causes and effects. Experiments show that the proposed model reduces Current Action L1 loss by 21.7% and Next Actions L1 loss by 18.4% compared to the baseline, demonstrating improved action prediction accuracy and explainability under adversarial conditions.

---

### Solving Parallel Machine Scheduling With Precedences and Cumulative Resource Constraints With Calendars

Authors: Christoph Einspieler, Matthias Horn, Marie-Louise Lackner, Patrick Malik, Nysret Musliu, Felix Winter
Published: 2025-12-05
Categories: cs.AI, math.OC
arXiv: https://arxiv.org/abs/2512.11864v1

The task of finding efficient production schedules for parallel machines is a challenge that arises in most industrial manufacturing domains. There is a large potential to minimize production costs through automated scheduling techniques, due to the large-scale requirements of modern factories. In the past, solution approaches have been studied for many machine scheduling variations, where even basic variants have been shown to be NP-hard. However, in today's real-life production environments, additional complex precedence constraints and resource restrictions with calendars arise that must be fulfilled. These additional constraints cannot be tackled efficiently by existing solution techniques. Thus, there is a strong need to develop and analyze automated methods that can solve such real-life parallel machine scheduling scenarios. In this work, we introduce a novel variant of parallel machine scheduling with job precedences and calendar-based cumulative resource constraints that arises in real-life industrial use cases. A constraint modeling approach is proposed as an exact solution method for small scheduling scenarios together with state-of-the-art constraint-solving technology. Further, we propose a construction heuristic as well as a tailored metaheuristic using local search to efficiently tackle large-scale problem instances. This metaheuristic approach has been deployed and is currently being used in an industrial setting.

---

### Expert Assessment: The Systemic Environmental Risks of Artficial Intelligence

Authors: Julian Schön, Lena Hoffmann, Nikolas Becker
Published: 2025-12-05
Categories: cs.CY, cs.AI
arXiv: https://arxiv.org/abs/2512.11863v1

Artificial intelligence (AI) is often presented as a key tool for addressing societal challenges, such as climate change. At the same time, AI's environmental footprint is expanding increasingly. This report describes the systemic environmental risks of artificial intelligence, in particular, moving beyond direct impacts such as energy and water usage. Systemic environmental risks of AI are emergent, cross-sector harms to climate, biodiversity, freshwater, and broader socioecological systems that arise primarily from AI's integration into social, economic, and physical infrastructures, rather than its direct resource use, and that propagate through feedbacks, yielding nonlinear, inequitable, and potentially irreversible impacts. While these risks are emergent and quantification is uncertain, this report aims to provide an overview of systemic environmental risks. Drawing on a narrative literature review, we propose a three-level framework that operationalizes systemic risk analysis. The framework identifies the structural conditions that shape AI development, the risk amplification mechanisms that propagate environmental harm, and the impacts that manifest as observable ecological and social consequences. We illustrate the framework in expert-interview-based case studies across agriculture and biodiversity, oil and gas, and waste management.

---

### Hierarchical Task Offloading and Trajectory Optimization in Low-Altitude Intelligent Networks Via Auction and Diffusion-based MARL

Authors: Jiahao You, Ziye Jia, Can Cui, Chao Dong, Qihui Wu, Zhu Han
Published: 2025-12-05
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2512.11862v1

The low-altitude intelligent networks (LAINs) emerge as a promising architecture for delivering low-latency and energy-efficient edge intelligence in dynamic and infrastructure-limited environments. By integrating unmanned aerial vehicles (UAVs), aerial base stations, and terrestrial base stations, LAINs can support mission-critical applications such as disaster response, environmental monitoring, and real-time sensing. However, these systems face key challenges, including energy-constrained UAVs, stochastic task arrivals, and heterogeneous computing resources. To address these issues, we propose an integrated air-ground collaborative network and formulate a time-dependent integer nonlinear programming problem that jointly optimizes UAV trajectory planning and task offloading decisions. The problem is challenging to solve due to temporal coupling among decision variables. Therefore, we design a hierarchical learning framework with two timescales. At the large timescale, a Vickrey-Clarke-Groves auction mechanism enables the energy-aware and incentive-compatible trajectory assignment. At the small timescale, we propose the diffusion-heterogeneous-agent proximal policy optimization, a generative multi-agent reinforcement learning algorithm that embeds latent diffusion models into actor networks. Each UAV samples actions from a Gaussian prior and refines them via observation-conditioned denoising, enhancing adaptability and policy diversity. Extensive simulations show that our framework outperforms baselines in energy efficiency, task success rate, and convergence performance.

---

### An Operator-Consistent Graph Neural Network for Learning Diffusion Dynamics on Irregular Meshes

Authors: Yuelian Li, Andrew Rushing Hands
Published: 2025-12-05
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2512.11860v1

Classical numerical methods solve partial differential equations (PDEs) efficiently on regular meshes, but many of them become unstable on irregular domains. In practice, multiphysics interactions such as diffusion, damage, and healing often take place on irregular meshes. We develop an operator-consistent graph neural network (OCGNN-PINN) that approximates PDE evolution under physics-informed constraints. It couples node-edge message passing with a consistency loss enforcing the gradient-divergence relation through the graph incidence matrix, ensuring that discrete node and edge dynamics remain structurally coupled during temporal rollout. We evaluate the model on diffusion processes over physically driven evolving meshes and real-world scanned surfaces. The results show improved temporal stability and prediction accuracy compared with graph convolutional and multilayer perceptron baselines, approaching the performance of Crank-Nicolson solvers on unstructured domains.

---

### Generative Stochastic Optimal Transport: Guided Harmonic Path-Integral Diffusion

Authors: Michael Chertkov
Published: 2025-12-05
Categories: cs.LG, cond-mat.stat-mech, cs.AI, eess.SY, stat.ML
arXiv: https://arxiv.org/abs/2512.11859v1

We introduce Guided Harmonic Path-Integral Diffusion (GH-PID), a linearly-solvable framework for guided Stochastic Optimal Transport (SOT) with a hard terminal distribution and soft, application-driven path costs. A low-dimensional guidance protocol shapes the trajectory ensemble while preserving analytic structure: the forward and backward Kolmogorov equations remain linear, the optimal score admits an explicit Green-function ratio, and Gaussian-Mixture Model (GMM) terminal laws yield closed-form expressions. This enables stable sampling and differentiable protocol learning under exact terminal matching. We develop guidance-centric diagnostics -- path cost, centerline adherence, variance flow, and drift effort -- that make GH-PID an interpretable variational ansatz for empirical SOT. Three navigation scenarios illustrated in 2D: (i) Case A: hand-crafted protocols revealing how geometry and stiffness shape lag, curvature effects, and mode evolution; (ii) Case B: single-task protocol learning, where a PWC centerline is optimized to minimize integrated cost; (iii) Case C: multi-expert fusion, in which a commander reconciles competing expert/teacher trajectories and terminal beliefs through an exact product-of-experts law and learns a consensus protocol. Across all settings, GH-PID generates geometry-aware, trust-aware trajectories that satisfy the prescribed terminal distribution while systematically reducing integrated cost.

---

### Adaptive Path Integral Diffusion: AdaPID

Authors: Michael Chertkov, Hamidreza Behjoo
Published: 2025-12-05
Categories: cs.LG, cond-mat.stat-mech, cs.AI, eess.SY, stat.ML
arXiv: https://arxiv.org/abs/2512.11858v1

Diffusion-based samplers -- Score Based Diffusions, Bridge Diffusions and Path Integral Diffusions -- match a target at terminal time, but the real leverage comes from choosing the schedule that governs the intermediate-time dynamics. We develop a path-wise schedule -- selection gramework for Harmonic PID with a time-varying stiffness, exploiting Piece-Wise-Constant(PWC) parametrizations and a simple hierarchical refinement. We introduce schedule-sensitive Quality-of-Sampling (QoS) diagnostics. Assuming a Gaussian-Mixture (GM) target, we retain closed-form Green functions' ration and numerically stable, Neural-Network free oracles for predicted-state maps and score. Experiments in 2D show that QoS driven PWC schedules consistently improve early-exit fidelity, tail accuracy, conditioning of the dynamics, and speciation (label-selection) timing at fixed integration budgets.

---

### TopicProphet: Prophesies on Temporal Topic Trends and Stocks

Authors: Olivia Kim
Published: 2025-12-05
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2512.11857v1

Stocks can't be predicted. Despite many hopes, this premise held itself true for many years due to the nature of quantitative stock data lacking causal logic along with rapid market changes hindering accumulation of significant data for training models. To undertake this matter, we propose a novel framework, TopicProphet, to analyze historical eras that share similar public sentiment trends and historical background. Our research deviates from previous studies that identified impacts of keywords and sentiments - we expand on that method by a sequence of topic modeling, temporal analysis, breakpoint detection and segment optimization to detect the optimal time period for training. This results in improving predictions by providing the model with nuanced patterns that occur from that era's socioeconomic and political status while also resolving the shortage of pertinent stock data to train on. Through extensive analysis, we conclude that TopicProphet produces improved outcomes compared to the state-of-the-art methods in capturing the optimal training data for forecasting financial percentage changes.

---

### GCoDE: Efficient Device-Edge Co-Inference for GNNs via Architecture-Mapping Co-Search

Authors: Ao Zhou, Jianlei Yang, Tong Qiao, Yingjie Qi, Zhi Yang, Weisheng Zhao, Chunming Hu
Published: 2025-12-05
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2512.11856v1

Graph Neural Networks (GNNs) have emerged as the state-of-the-art graph learning method. However, achieving efficient GNN inference on edge devices poses significant challenges, limiting their application in real-world edge scenarios. This is due to the high computational cost of GNNs and limited hardware resources on edge devices, which prevent GNN inference from meeting real-time and energy requirements. As an emerging paradigm, device-edge co-inference shows potential for improving inference efficiency and reducing energy consumption on edge devices. Despite its potential, research on GNN device-edge co-inference remains scarce, and our findings show that traditional model partitioning methods are ineffective for GNNs. To address this, we propose GCoDE, the first automatic framework for GNN architecture-mapping Co-design and deployment on Device-Edge hierarchies. By abstracting the device communication process into an explicit operation, GCoDE fuses the architecture and mapping scheme in a unified design space for joint optimization. Additionally, GCoDE's system performance awareness enables effective evaluation of architecture efficiency across diverse heterogeneous systems. By analyzing the energy consumption of various GNN operations, GCoDE introduces an energy prediction method that improves energy assessment accuracy and identifies energy-efficient solutions. Using a constraint-based random search strategy, GCoDE identifies the optimal solution in 1.5 hours, balancing accuracy and efficiency. Moreover, the integrated co-inference engine in GCoDE enables efficient deployment and execution of GNN co-inference. Experimental results show that GCoDE can achieve up to 44.9x speedup and 98.2% energy reduction compared to existing approaches across diverse applications and system configurations.

---

### Achieving Approximate Symmetry Is Exponentially Easier than Exact Symmetry

Authors: Behrooz Tahmasebi, Melanie Weber
Published: 2025-12-05
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2512.11855v2

Enforcing exact symmetry in machine learning models often yields significant gains in scientific applications, serving as a powerful inductive bias. However, recent work suggests that relying on approximate symmetry can offer greater flexibility and robustness. Despite promising empirical evidence, there has been little theoretical understanding, and in particular, a direct comparison between exact and approximate symmetry is missing from the literature. In this paper, we initiate this study by asking: What is the cost of enforcing exact versus approximate symmetry? To address this question, we introduce averaging complexity, a framework for quantifying the cost of enforcing symmetry via averaging. Our main result is an exponential separation: under standard conditions, exact symmetry requires linear averaging complexity, whereas approximate symmetry can be attained with only logarithmic complexity in the group size. To the best of our knowledge, this provides the first theoretical separation of these two cases, formally justifying why approximate symmetry may be preferable in practice. Beyond this, our tools and techniques may be of independent interest for the broader study of symmetries in machine learning.

---

### Rep Smarter, Not Harder: AI Hypertrophy Coaching with Wearable Sensors and Edge Neural Networks

Authors: Grant King, Musa Azeem, Savannah Noblitt, Ramtin Zand, Homayoun Valafar
Published: 2025-12-05
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2512.11854v1

Optimizing resistance training for hypertrophy requires balancing proximity to muscular failure, often quantified by Repetitions in Reserve (RiR), with fatigue management. However, subjective RiR assessment is unreliable, leading to suboptimal training stimuli or excessive fatigue. This paper introduces a novel system for real-time feedback on near-failure states (RiR $\le$ 2) during resistance exercise using only a single wrist-mounted Inertial Measurement Unit (IMU). We propose a two-stage pipeline suitable for edge deployment: first, a ResNet-based model segments repetitions from the 6-axis IMU data in real-time. Second, features derived from this segmentation, alongside direct convolutional features and historical context captured by an LSTM, are used by a classification model to identify exercise windows corresponding to near-failure states. Using a newly collected dataset from 13 diverse participants performing preacher curls to failure (631 total reps), our segmentation model achieved an F1 score of 0.83, and the near-failure classifier achieved an F1 score of 0.82 under simulated real-time evaluation conditions (1.6 Hz inference rate). Deployment on a Raspberry Pi 5 yielded an average inference latency of 112 ms, and on an iPhone 16 yielded 23.5 ms, confirming the feasibility for edge computation. This work demonstrates a practical approach for objective, real-time training intensity feedback using minimal hardware, paving the way for accessible AI-driven hypertrophy coaching tools that help users manage intensity and fatigue effectively.

---

### Evolving Deep Learning Optimizers

Authors: Mitchell Marfinetz
Published: 2025-12-05
Categories: cs.NE, cs.LG
arXiv: https://arxiv.org/abs/2512.11853v1

We present a genetic algorithm framework for automatically discovering deep learning optimization algorithms. Our approach encodes optimizers as genomes that specify combinations of primitive update terms (gradient, momentum, RMS normalization, Adam-style adaptive terms, and sign-based updates) along with hyperparameters and scheduling options. Through evolutionary search over 50 generations with a population of 50 individuals, evaluated across multiple vision tasks, we discover an evolved optimizer that outperforms Adam by 2.6% in aggregate fitness and achieves a 7.7% relative improvement on CIFAR-10. The evolved optimizer combines sign-based gradient terms with adaptive moment estimation, uses lower momentum coefficients than Adam ($β_1$=0.86, $β_2$=0.94), and notably disables bias correction while enabling learning rate warmup and cosine decay. Our results demonstrate that evolutionary search can discover competitive optimization algorithms and reveal design principles that differ from hand-crafted optimizers. Code is available at https://github.com/mmarfinetz/evo-optimizer.

---

### Marti-5: A Mathematical Model of "Self in the World" as a First Step Toward Self-Awareness

Authors: Igor Pivovarov, Sergey Shumsky
Published: 2025-12-05
Categories: q-bio.NC, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2512.10985v1

The existence of 'what' and 'where' pathways of information processing in the brain was proposed almost 30 years ago, but there is still a lack of a clear mathematical model that could show how these pathways work together. We propose a biologically inspired mathematical model that uses this idea to identify and separate the self from the environment and then build and use a self-model for better predictions. This is a model of neocortical columns governed by the basal ganglia to make predictions and choose the next action, where some columns act as 'what' columns and others act as 'where' columns. Based on this model, we present a reinforcement learning agent that learns purposeful behavior in a virtual environment. We evaluate the agent on the Atari games Pong and Breakout, where it successfully learns to play. We conclude that the ability to separate the self from the environment gives advantages to the agent and therefore such a model could appear in living organisms during evolution. We propose Self-Awareness Principle 1: the ability to separate the self from the world is a necessary but insufficient condition for self-awareness.

---

### HSCP: A Two-Stage Spectral Clustering Framework for Resource-Constrained UAV Identification

Authors: Maoyu Wang, Yao Lu, Bo Zhou, Zhuangzhi Chen, Yun Lin, Qi Xuan, Guan Gui
Published: 2025-12-05
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2512.08983v1

With the rapid development of Unmanned Aerial Vehicles (UAVs) and the increasing complexity of low-altitude security threats, traditional UAV identification methods struggle to extract reliable signal features and meet real-time requirements in complex environments. Recently, deep learning based Radio Frequency Fingerprint Identification (RFFI) approaches have greatly improved recognition accuracy. However, their large model sizes and high computational demands hinder deployment on resource-constrained edge devices. While model pruning offers a general solution for complexity reduction, existing weight, channel, and layer pruning techniques struggle to concurrently optimize compression rate, hardware acceleration, and recognition accuracy. To this end, in this paper, we introduce HSCP, a Hierarchical Spectral Clustering Pruning framework that combines layer pruning with channel pruning to achieve extreme compression, high performance, and efficient inference. In the first stage, HSCP employs spectral clustering guided by Centered Kernel Alignment (CKA) to identify and remove redundant layers. Subsequently, the same strategy is applied to the channel dimension to eliminate a finer redundancy. To ensure robustness, we further employ a noise-robust fine-tuning strategy. Experiments on the UAV-M100 benchmark demonstrate that HSCP outperforms existing channel and layer pruning methods. Specifically, HSCP achieves $86.39\%$ parameter reduction and $84.44\%$ FLOPs reduction on ResNet18 while improving accuracy by $1.49\%$ compared to the unpruned baseline, and maintains superior robustness even in low signal-to-noise ratio environments.

---

### Consist-Retinex: One-Step Noise-Emphasized Consistency Training Accelerates High-Quality Retinex Enhancement

Authors: Jian Xu, Wei Chen, Shigui Li, Delu Zeng, John Paisley, Qibin Zhao
Published: 2025-12-05
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2512.08982v4

Retinex-based low-light image enhancement benefits from separating reflectance and illumination, yet recent generative approaches often rely on iterative sampling and are difficult to deploy under strict latency budgets. Consistency models offer a natural route to one-step restoration, but direct adaptation to Retinex-factorized enhancement is unstable: one-step inference is evaluated at the high-noise endpoint, whereas standard training schedules provide little supervision there, and temporal self-consistency alone does not determine the correct conditional target. We propose Consist-Retinex, which first uses a Retinex Transformer Decomposition Network (TDN) to obtain paired reflectance and illumination maps, then trains two conditional consistency models with a Retinex-aware dual objective and adaptive noise-emphasized fixed-point sampling. The dual objective combines trajectory consistency with paired ground-truth component alignment, while the sampling rule concentrates supervision near the inference endpoint without discarding full-range noise coverage. We further provide an endpoint error bound, an anchoring-propagation result, and a high-noise sample-allocation analysis that explain why endpoint supervision and temporal consistency are complementary for one-step Retinex enhancement. Experiments on paired and unpaired low-light benchmarks show that Consist-Retinex obtains the best VE-LOL-L scores among the compared methods under one-step inference and remains competitive on LOL, with substantially reduced sampling and consistency-stage training cost in the reported setup.

---

### Mitigating Bias with Words: Inducing Demographic Ambiguity in Face Recognition Templates by Text Encoding

Authors: Tahar Chettaoui, Naser Damer, Fadi Boutros
Published: 2025-12-05
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2512.08981v1

Face recognition (FR) systems are often prone to demographic biases, partially due to the entanglement of demographic-specific information with identity-relevant features in facial embeddings. This bias is extremely critical in large multicultural cities, especially where biometrics play a major role in smart city infrastructure. The entanglement can cause demographic attributes to overshadow identity cues in the embedding space, resulting in disparities in verification performance across different demographic groups. To address this issue, we propose a novel strategy, Unified Text-Image Embedding (UTIE), which aims to induce demographic ambiguity in face embeddings by enriching them with information related to other demographic groups. This encourages face embeddings to emphasize identity-relevant features and thus promotes fairer verification performance across groups. UTIE leverages the zero-shot capabilities and cross-modal semantic alignment of Vision-Language Models (VLMs). Given that VLMs are naturally trained to align visual and textual representations, we enrich the facial embeddings of each demographic group with text-derived demographic features extracted from other demographic groups. This encourages a more neutral representation in terms of demographic attributes. We evaluate UTIE using three VLMs, CLIP, OpenCLIP, and SigLIP, on two widely used benchmarks, RFW and BFW, designed to assess bias in FR. Experimental results show that UTIE consistently reduces bias metrics while maintaining, or even improving in several cases, the face verification accuracy.

---

### Training Multi-Image Vision Agents via End2End Reinforcement Learning

Authors: Chengqi Dong, Chuhuai Yue, Hang He, Rongge Mao, Fenghe Tang, S Kevin Zhou, Zekun Xu, Xiaohan Wang, Jiajun Chai, Guojun Yin
Published: 2025-12-05
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2512.08980v3

Recent VLM-based agents aim to replicate OpenAI O3's "thinking with images" via tool use, yet most open-source methods restrict inputs to a single image, limiting their applicability to real-world multi-image QA tasks. To address this gap, we propose IMAgent, an open-source visual agent trained with end-to-end reinforcement learning for fine-grained single/multi-image reasoning. During inference, VLMs tend to gradually neglect visual inputs; to mitigate this issue, we design two dedicated tools for visual reflection and verification, enabling the model to actively refocus attention on image content. Beyond that, we, for the first time, reveal how tool usage enhances agent performance from an attention perspective. Equipped with a carefully designed two-layer motion trajectory masking strategy and tool-use reward gain, IMAgent acquires an effective tool-use paradigm through pure reinforcement learning, eliminating the need for costly supervised fine-tuning data. To further unleash the inherent tool-usage potential of the base VLM and fill data gaps, we construct a challenging, visually enriched multi-image QA dataset via multi-agent system. Extensive experiments validate that IMAgent achieves SOTA performance across mainstream single and multi-image benchmarks, and our in-depth analysis offers actionable insights for the community. Code and data will be released soon.

---

### What Happens When: Learning Temporal Orders of Events in Videos

Authors: Daechul Ahn, Yura Choi, Hyeonbeom Choi, Seongwon Cho, San Kim, Jonghyun Choi
Published: 2025-12-05
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2512.08979v1

Video Large Multimodal Models (VLMMs) have shown impressive performance in video understanding, yet their ability to accurately capture the temporal order of multiple events remains underexplored. We interestingly observe that, even when video frames are scrambled, models perform very well on the existing benchmarks by comprehensive experiments. This implies that VLMMs may not necessarily rely on accurate sequential processing of visual events, but instead depend on prior knowledge of typical scenarios to answer the question. To benchmark temporal understanding capabilities in VLMMs, we propose VECTOR, designed to explicitly assess a model's ability to identify the temporal order of events. On this benchmark, we observe that various VLMMs often fail to understand the orders of events. To address this, we propose MECOT (Multi-Event instruction fine-tuning with Chain-of-Thought), which (1) trains models on detailed, event-by-event video descriptions and (2) using chain-of-thought prompts at inference to enhance temporal awareness. MECOT outperforms prior arts on VECTOR as well as improving performance on existing video benchmarks, implying effectiveness of temporal understanding. We release our code, model and datasets.

---

### The Theory of Strategic Evolution: Games with Endogenous Players and the Seven Laws of Strategic Replicators

Authors: Kevin Vallier
Published: 2025-12-05
Categories: cs.GT, cs.AI, cs.CY, cs.MA, econ.TH
arXiv: https://arxiv.org/abs/2512.07901v4

Von Neumann founded both game theory and the theory of self-reproducing automata, but the two programs never merged. Rational players do not control their replication, and replicators do not choose strategically. Contemporary AI systems expose this gap: they optimize objectives, yet the population of AI systems is not fixed but expands and contracts based on performance. When capital can spawn capital, we need a theory that captures both rationality and replication. The Theory of Strategic Evolution analyzes strategic replicators: entities that optimize under resource constraints and spawn copies of themselves. The framework is organized around Seven Laws: 1. Strategic Selection: Mean fitness serves as a Lyapunov function; dominated types are eliminated. 2. ESDI Characterization: Equilibria exist, are generically finite, and satisfy Nash-KKT-LP equivalence. 3. H-$γ$ Stability: Multi-level systems are stable iff the spectral radius $ρ(Γ) < 1$. 4. G$\infty$ Closure: A unique maximal class of safe modifications exists and is closed under composition. 5. Constitutional Duality: Shadow prices implement any frontier allocation; welfare theorems hold. 6. Alignment Impossibility: Full reachability destroys Lyapunov structure; alignment requires bounded modification. 7. Hopf Transition: At critical coupling, systems undergo supercritical bifurcation producing limit cycles. Applications range from AI deployment dynamics to institutional design. The framework shows why "personality engineering" fails under selection pressure and identifies constitutional constraints necessary for stable alignment. The Lean 4 formalization verifies every result with zero custom axioms. The framework generates empirical predictions that remain untested. The paper and its repository are therefore a prototype kernel: an instrument for studying strategic replicators, not a set of results about the world.

---

### MARINE: Theoretical Optimization and Design for Multi-Agent Recursive IN-context Enhancement

Authors: Hongwei Zhang, Ji Lu, Yongsheng Du, Yanqin Gao, Lingjun Huang, Baoli Wang, Fang Tan, Peng Zou
Published: 2025-12-05
Categories: cs.MA, cs.AI
arXiv: https://arxiv.org/abs/2512.07898v1

Large Language Model (LLM)-based agents demonstrate advanced reasoning capabilities, yet practical constraints frequently limit outputs to single responses, leaving significant performance potential unrealized. This paper introduces MARINE (Multi-Agent Recursive IN-context Enhancement), a theoretically grounded framework that reconceptualizes test-time reasoning as iterative refinement of a persistent reference trajectory, fundamentally departing from conventional one-shot or multi-sample paradigms. The MARINE refinement operator systematically converts a base model's pass@N capabilities into near-optimal pass@1 performance. Rigorous theoretical analysis establishes that minimal feasible batches maximize expected performance gains under fixed invocation budgets, while logarithmically growing batch schedules ensure continuous improvement without computational constraints. Comprehensive evaluation on the BrowserComp-ZH benchmark demonstrates state-of-the-art results, with a 685B-parameter implementation achieving 46.0% pass@1 accuracy. Meanwhile, MARINE establishes a new paradigm for parameter-efficient reasoning: an 80B-parameter model augmented with MARINE matches the performance of standalone 1000B-parameter agents, reducing parameter requirements by over an order of magnitude. Notably, within a fixed computational budget, the proposed MARINE delivers higher-quality samples to alignment and optimization processes than traditional sampling-and-ranking strategies. Consequently, it has great potential to boost post-training efficiency.

---

### Average-reward reinforcement learning in semi-Markov decision processes via relative value iteration

Authors: Huizhen Yu, Yi Wan, Richard S. Sutton
Published: 2025-12-05
Categories: cs.LG, math.OC
arXiv: https://arxiv.org/abs/2512.06218v1

This paper applies the authors' recent results on asynchronous stochastic approximation (SA) in the Borkar-Meyn framework to reinforcement learning in average-reward semi-Markov decision processes (SMDPs). We establish the convergence of an asynchronous SA analogue of Schweitzer's classical relative value iteration algorithm, RVI Q-learning, for finite-space, weakly communicating SMDPs. In particular, we show that the algorithm converges almost surely to a compact, connected subset of solutions to the average-reward optimality equation, with convergence to a unique, sample path-dependent solution under additional stepsize and asynchrony conditions. Moreover, to make full use of the SA framework, we introduce new monotonicity conditions for estimating the optimal reward rate in RVI Q-learning. These conditions substantially expand the previously considered algorithmic framework and are addressed through novel arguments in the stability and convergence analysis of RVI Q-learning.

---

### A Broader View on Clustering under Cluster-Aware Norm Objectives

Authors: Martin G. Herold, Evangelos Kipouridis, Joachim Spoerhase
Published: 2025-12-05
Categories: cs.DS, cs.LG
arXiv: https://arxiv.org/abs/2512.06211v2

We revisit the $(f,g)$-clustering problem that we introduced in a recent work (SODA'25), and which subsumes fundamental clustering problems such as $k$-Center, $k$-Median, Min-Sum of Radii, and Min-Load $k$-Clustering. This problem assigns each of the $k$ clusters a cost determined by the monotone, symmetric norm $f$ applied to the vector distances in the cluster, and aims at minimizing the norm $g$ applied to the vector of cluster costs. Previously, we focused on certain special cases for which we designed constant-factor approximation algorithms. Our bounds for more general settings left, however, large gaps to the known bounds for the basic problems they capture. In this work, we provide a clearer picture of the approximability of these more general settings. First, we design an $O(\log^2 n)$-approximation algorithm for $(f, L_{1})$-clustering for any $f$. This improves upon our previous $\widetilde{O}(\sqrt{n})$-approximation. Second, we provide an $O(k)$-approximation for the general $(f,g)$-clustering problem, which improves upon our previous $\widetilde{O}(\sqrt{kn})$-approximation algorithm and matches the best-known upper bound for Min-Load $k$-Clustering. We then design an approximation algorithm for $(f,g)$-clustering that interpolates, up to polylog factors, between the best known bounds for $k$-Center, $k$-Median, Min-Sum of Radii, Min-Load $k$-Clustering, (Top, $L_{1}$)-clustering, and $(L_{\infty},g)$-clustering based on a newly defined parameter of $f$ and $g$.

---

### Forests of Uncertaint(r)ees: Using tree-based ensembles to estimate probability distributions of future conflict

Authors: Daniel Mittermaier, Tobias Bohne, Martin Hofer, Daniel Racek
Published: 2025-12-05
Categories: stat.AP, cs.LG
arXiv: https://arxiv.org/abs/2512.06210v2

Predictions of fatalities from violent conflict on the PRIO-GRID-month (pgm) level are characterized by high levels of uncertainty, limiting their usefulness in practical applications. We discuss the two main sources of uncertainty for this prediction task, the nature of violent conflict and data limitations, embedding conflict prediction in the wider literature on uncertainty quantification in machine learning. Based on this, we develop a strategy to quantify uncertainty in conflict forecasting, shifting from traditional point predictions to full predictive distributions. Our approach combines multiple tree-based classifiers and distributional regressors in a custom AutoML setup, estimating distributions for each pgm individually. We also test the integration of regional models in spatial ensembles as a potential avenue to reduce uncertainty by lowering data requirements and accounting for systematic differences between conflict contexts. The models are able to consistently outperform a suite of benchmarks derived from conflict history in predictions up to one year in advance. Marginal differences in model-wide metrics emphasize the need to understand their behavior for a given prediction problem, in this case characterized by extremely high zero-inflatedness. Adressing this, we compliment our evaluation with a simulation experiment, which demonstrates that our models reflect meaningful performance improvements, which can be traced back to conflict-affected regions. Lastly, we show that the integration of regional models does not decrease performance, opening avenues to integrate additional data sources in the future.

---

### SparsePixels: Efficient Convolution for Sparse Data on FPGAs

Authors: Ho Fung Tsoi, Dylan Rankin, Vladimir Loncar, Philip Harris
Published: 2025-12-05
Categories: cs.AR, cs.LG, hep-ex
arXiv: https://arxiv.org/abs/2512.06208v4

Inference of standard convolutional neural networks (CNNs) on FPGAs often incurs high latency and a long initiation interval due to the deep nested loops required to densely convolve every input pixel regardless of its feature value. However, input features can be spatially sparse in some image data, where semantic information may occupy only a small fraction of the pixels and most computation would be wasted on empty regions. In this work, we introduce SparsePixels, a framework that implements sparse convolution on FPGAs by selectively retaining and computing on a small subset of active input pixels while ignoring the rest, reducing the per-layer compute from $O(HWK^2C_{\text{in}}C_{\text{out}})$ to $O(n^2C_{\text{in}}C_{\text{out}})$ and the storage from $O(HWC)$ to $O(nC)$ when $n^2\ll HW$. Because computation always runs over a single pre-specified pixel budget that is frozen at synthesis, the inference latency is data-independent and constant at runtime. We show that, for identifying neutrino interactions in naturally sparse LArTPC images with 4k pixels, a standard CNN with a compact size of 4k parameters incurs an inference latency of 48.665 $μ$s on an FPGA, whereas a sparse CNN of the same base architecture, computing on less than 1% of the input pixels, achieves a $\times 73$ speedup to 0.665 $μ$s with resource utilization well within on-chip budgets, trading only a small percent-level performance loss. This work aims to benefit future algorithm development for efficient data readout in modern experiments with strict latency requirements of microseconds or below.

---

### The MICCAI Federated Tumor Segmentation (FeTS) Challenge 2024: Efficient and Robust Aggregation Methods for Federated Learning

Authors: Akis Linardos, Sarthak Pati, Ujjwal Baid, Brandon Edwards, Patrick Foley, Kevin Ta, Verena Chung, Micah Sheller, Muhammad Irfan Khan, Mojtaba Jafaritadi, Elina Kontio, Suleiman Khan et al.
Published: 2025-12-05
Categories: cs.CV, cs.LG
arXiv: https://arxiv.org/abs/2512.06206v1

We present the design and results of the MICCAI Federated Tumor Segmentation (FeTS) Challenge 2024, which focuses on federated learning (FL) for glioma sub-region segmentation in multi-parametric MRI and evaluates new weight aggregation methods aimed at improving robustness and efficiency. Six participating teams were evaluated using a standardized FL setup and a multi-institutional dataset derived from the BraTS glioma benchmark, consisting of 1,251 training cases, 219 validation cases, and 570 hidden test cases with segmentations for enhancing tumor (ET), tumor core (TC), and whole tumor (WT). Teams were ranked using a cumulative scoring system that considered both segmentation performance, measured by Dice Similarity Coefficient (DSC) and the 95th percentile Hausdorff Distance (HD95), and communication efficiency assessed through the convergence score. A PID-controller-based method achieved the top overall ranking, obtaining mean DSC values of 0.733, 0.761, and 0.751 for ET, TC, and WT, respectively, with corresponding HD95 values of 33.922 mm, 33.623 mm, and 32.309 mm, while also demonstrating the highest communication efficiency with a convergence score of 0.764. These findings advance the state of federated learning for medical imaging, surpassing top-performing methods from previous challenge iterations and highlighting PID controllers as effective mechanisms for stabilizing and optimizing weight aggregation in FL. The challenge code is available at https://github.com/FeTS-AI/Challenge.

---

### A framework for auditing grounding claims

Authors: Daniel Quigley, Eric Maynard
Published: 2025-12-05
Categories: cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2512.06205v3

The symbol grounding problem asks how a token such as cat can be about cats. We propose a framework for auditing grounding claims against a declared semantic standard. The audit reports measurements and evidence, with overall verdicts conditional on explicit acceptance criteria. Its profiles assess accuracy, robustness, and composition alongside evidence about how the system acquired its mechanisms, how they contribute to performance, and why they were retained. In a toy gridworld, an agent interprets individual symbols accurately but fails a withheld combination. Composing its interpretations by the declared rule would succeed. This comparison identifies a departure from the composition rule within the observed failure. Both this audit and a pilot on pretrained word vectors provide evidence that a designated mechanism contributes to present performance. Whether that contribution explains its retention remains uncertified. The framework evaluates the evidence for grounding claims; candidate accounts remain responsible for explaining how meaning emerges.

---

### Quantifying Memory Use in Reinforcement Learning with Temporal Range

Authors: Rodney Lafuente-Mercado, Daniela Rus, T. Konstantin Rusch
Published: 2025-12-05
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2512.06204v1

How much does a trained RL policy actually use its past observations? We propose \emph{Temporal Range}, a model-agnostic metric that treats first-order sensitivities of multiple vector outputs across a temporal window to the input sequence as a temporal influence profile and summarizes it by the magnitude-weighted average lag. Temporal Range is computed via reverse-mode automatic differentiation from the Jacobian blocks $\partial y_s/\partial x_t\in\mathbb{R}^{c\times d}$ averaged over final timesteps $s\in\{t+1,\dots,T\}$ and is well-characterized in the linear setting by a small set of natural axioms. Across diagnostic and control tasks (POPGym; flicker/occlusion; Copy-$k$) and architectures (MLPs, RNNs, SSMs), Temporal Range (i) remains small in fully observed control, (ii) scales with the task's ground-truth lag in Copy-$k$, and (iii) aligns with the minimum history window required for near-optimal return as confirmed by window ablations. We also report Temporal Range for a compact Long Expressive Memory (LEM) policy trained on the task, using it as a proxy readout of task-level memory. Our axiomatic treatment draws on recent work on range measures, specialized here to temporal lag and extended to vector-valued outputs in the RL setting. Temporal Range thus offers a practical per-sequence readout of memory dependence for comparing agents and environments and for selecting the shortest sufficient context.

---

### K2-V2: A 360-Open, Reasoning-Enhanced LLM

Authors: K2 Team, Zhengzhong Liu, Liping Tang, Linghao Jin, Haonan Li, Nikhil Ranjan, Desai Fan, Shaurya Rohatgi, Richard Fan, Omkar Pangarkar, Huijuan Wang, Zhoujun Cheng et al.
Published: 2025-12-05
Categories: cs.LG
arXiv: https://arxiv.org/abs/2512.06201v3

We introduce K2-V2, a 360-open LLM built from scratch as a superior base for reasoning adaptation, in addition to functions such as conversation and knowledge retrieval from general LLMs. It stands as the strongest fully open model, rivals open-weight leaders in its size class, outperforms Qwen2.5-72B and approaches the performance of Qwen3-235B. We actively infuse domain knowledge, reasoning, long-context, and tool use throughout the training process. This explicitly prepares the model for complex reasoning tasks. We demonstrate this potential using simple supervised fine-tuning, establishing a strong baseline that indicates significant headroom for advanced alignment. By releasing the full training history and data composition, we maximize the effectiveness of continuous training, a key open source production scenario. We release the model weights and signature LLM360 artifacts, such as complete training data, to empower the community with a capable, reasoning-centric foundation.

---

### How Should We Evaluate Data Deletion in Graph-Based ANN Indexes?

Authors: Tomohiro Yamashita, Daichi Amagata, Yusuke Matsui
Published: 2025-12-05
Categories: cs.LG
arXiv: https://arxiv.org/abs/2512.06200v1

Approximate Nearest Neighbor Search (ANNS) has recently gained significant attention due to its many applications, such as Retrieval-Augmented Generation. Such applications require ANNS algorithms that support dynamic data, so the ANNS problem on dynamic data has attracted considerable interest. However, a comprehensive evaluation methodology for data deletion in ANNS has yet to be established. This study proposes an experimental framework and comprehensive evaluation metrics to assess the efficiency of data deletion for ANNS indexes under practical use cases. Specifically, we categorize data deletion methods in graph-based ANNS into three approaches and formalize them mathematically. The performance is assessed in terms of accuracy, query speed, and other relevant metrics. Finally, we apply the proposed evaluation framework to Hierarchical Navigable Small World, one of the state-of-the-art ANNS methods, to analyze the effects of data deletion, and propose Deletion Control, a method which dynamically selects the appropriate deletion method under a required search accuracy.

---

### ARCANE: A Multi-Agent Framework for Interpretable and Configurable Alignment

Authors: Charlie Masters, Marta Grześkiewicz, Stefano V. Albrecht
Published: 2025-12-05
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2512.06196v1

As agents based on large language models are increasingly deployed to long-horizon tasks, maintaining their alignment with stakeholder preferences becomes critical. Effective alignment in such settings requires reward models that are interpretable so that stakeholders can understand and audit model objectives. Moreover, reward models must be capable of steering agents at interaction time, allowing preference shifts to be incorporated without retraining. We introduce ARCANE, a framework that frames alignment as a multi-agent collaboration problem that dynamically represents stakeholder preferences as natural-language rubrics: weighted sets of verifiable criteria that can be generated on-the-fly from task context. Inspired by utility theory, we formulate rubric learning as a reconstruction problem and apply a regularized Group-Sequence Policy Optimization (GSPO) procedure that balances interpretability, faithfulness, and computational efficiency. Using a corpus of 219 labeled rubrics derived from the GDPVal benchmark, we evaluate ARCANE on challenging tasks requiring multi-step reasoning and tool use. The learned rubrics produce compact, legible evaluations and enable configurable trade-offs (e.g., correctness vs. conciseness) without retraining. Our results show that rubric-based reward models offer a promising path toward interpretable, test-time adaptive alignment for complex, long-horizon AI systems.

---

### Do You Feel Comfortable? Detecting Hidden Conversational Escalation in AI Chatbots

Authors: Jihyung Park, Saleh Afroogh, David Atkinson, Junfeng Jiao
Published: 2025-12-05
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2512.06193v4

Large Language Models (LLM) are increasingly integrated into everyday interactions, serving not only as information assistants but also as emotional companions. Even in the absence of explicit toxicity, repeated emotional reinforcement or affective drift can gradually escalate distress in a form of \textit{implicit harm} that traditional toxicity filters fail to detect. Existing guardrail mechanisms often rely on external classifiers or clinical rubrics that may lag behind the nuanced, real-time dynamics of a developing conversation. To address this gap, we propose GAUGE (Guarding Affective Utterance Generation Escalation), logit-based framework for the real-time detection of hidden conversational escalation. GAUGE measures how an LLM's output probabilistically shifts the affective state of a dialogue.

---

### Multi-Modal Zero-Shot Prediction of Color Trajectories in Food Drying

Authors: Shichen Li, Ahmadreza Eslaminia, Chenhui Shao
Published: 2025-12-05
Categories: cs.CV, cs.AI, cs.LG, eess.IV
arXiv: https://arxiv.org/abs/2512.06190v1

Food drying is widely used to reduce moisture content, ensure safety, and extend shelf life. Color evolution of food samples is an important indicator of product quality in food drying. Although existing studies have examined color changes under different drying conditions, current approaches primarily rely on low-dimensional color features and cannot fully capture the complex, dynamic color trajectories of food samples. Moreover, existing modeling approaches lack the ability to generalize to unseen process conditions. To address these limitations, we develop a novel multi-modal color-trajectory prediction method that integrates high-dimensional temporal color information with drying process parameters to enable accurate and data-efficient color trajectory prediction. Under unseen drying conditions, the model attains RMSEs of 2.12 for cookie drying and 1.29 for apple drying, reducing errors by over 90% compared with baseline models. These experimental results demonstrate the model's superior accuracy, robustness, and broad applicability.

---