# Research Papers — 2026-01 · bundle page 155 of 189 > Full-text markdown of papers #7701–7750 of 9403 published in 2026-01, newest first. AI-readable mirrors of the full 9403-paper 2026-01 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-01/page/155 [← Previous bundle](https://aipod.dev/papers/archive/2026-01/page/154.md) [Next bundle →](https://aipod.dev/papers/archive/2026-01/page/156.md) ### Merging Triggers, Breaking Backdoors: Defensive Poisoning for Instruction-Tuned Language Models Authors: San Kim, Gary Geunbae Lee Published: 2026-01-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.04448v4 Large Language Models (LLMs) have greatly advanced Natural Language Processing (NLP), particularly through instruction tuning, which enables broad task generalization without additional fine-tuning. However, their reliance on large-scale datasets-often collected from human or web sources-makes them vulnerable to backdoor attacks, where adversaries poison a small subset of data to implant hidden behaviors. Despite this growing risk, defenses for instruction-tuned models remain underexplored. We propose MB-Defense (Merging & Breaking Defense Framework), a novel training pipeline that immunizes instruction-tuned LLMs against diverse backdoor threats. MB-Defense comprises two stages: (i) Defensive Poisoning, which merges attacker and defensive triggers into a unified backdoor representation, and (ii) Backdoor Neutralization, which breaks this representation through additional training to restore clean behavior. Extensive experiments across multiple LLMs show that MB-Defense substantially lowers attack success rates while preserving instruction-following ability. Our method offers a generalizable and data-efficient defense strategy, improving the robustness of instruction-tuned LLMs against unseen backdoor attacks. --- ### When Predictions Shape Reality: A Socio-Technical Synthesis of Performative Predictions in Machine Learning Authors: Gal Fybish, Teo Susnjak Published: 2026-01-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.04447v1 Machine learning models are increasingly used in high-stakes domains where their predictions can actively shape the environments in which they operate, a phenomenon known as performative prediction. This dynamic, in which the deployment of the model influences the very outcome it seeks to predict, can lead to unintended consequences, including feedback loops, performance issues, and significant societal risks. While the literature in the field has grown rapidly in recent years, a socio-technical synthesis that systemises the phenomenon concepts and provides practical guidance has been lacking. This Systematisation of Knowledge (SoK) addresses this gap by providing a comprehensive review of the literature on performative predictions. We provide an overview of the primary mechanisms through which performativity manifests, present a typology of associated risks, and survey the proposed solutions offered in the literature. Our primary contribution is the ``Performative Strength vs. Impact Matrix" assessment framework. This practical tool is designed to help practitioners assess the potential influence and severity of performativity on their deployed predictive models and select the appropriate level of algorithmic or human intervention. --- ### SpectraFormer: an Attention-Based Raman Unmixing Tool for Accessing the Graphene Buffer-Layer Signature on SiC Authors: Dmitriy Poteryayev, Pietro Novelli, Annalisa Coriolano, Riccardo Dettori, Valentina Tozzini, Fabio Beltram, Massimiliano Pontil, Antonio Rossi, Stiven Forti, Camilla Coletti Published: 2026-01-07 Categories: cond-mat.mtrl-sci, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.04445v1 Raman spectroscopy is a key tool for graphene characterization, yet its application to graphene grown on silicon carbide (SiC) is strongly limited by the intense and variable second-order Raman response of the substrate. This limitation is critical for buffer layer graphene, a semiconducting interfacial phase, whose vibrational signatures are overlapped with the SiC background and challenging to be reliably accessed using conventional reference-based subtraction, due to strong spatial and experimental variability of the substrate signal. Here we present SpectraFormer, a transformer-based deep learning model that reconstructs the SiC Raman substrate contribution directly from post-growth partially masked spectroscopic data without relying on explicit reference measurements. By learning global correlations across the entire Raman shift range, the model captures the statistical structure of the SiC background and enables accurate reconstruction of its contribution in mixed spectra. Subtraction of the reconstructed substrate signal reveals weak vibrational features associated with ZLG that are inaccessible through conventional analysis methods. The extracted spectra are validated by ab initio vibrational calculations, allowing assignment of the resolved features to specific modes and confirming their physical consistency. By leveraging a state-of-the-art attention-based deep learning architecture, this approach establishes a robust, reference-free framework for Raman analysis of graphene on SiC and provides a foundation, compatible with real-time data acquisition, to its integration into automated, closed-loop AI-assisted growth optimization. --- ### Large Language Models for Detecting Cyberattacks on Smart Grid Protective Relays Authors: Ahmad Mohammad Saber, Saeed Jafari, Zhengmao Ouyang, Paul Budnarain, Amr Youssef, Deepa Kundur Published: 2026-01-07 Categories: cs.CR, cs.LG, eess.SP arXiv: https://arxiv.org/abs/2601.04443v2 This paper presents a large language model (LLM)-based framework that adapts and fine-tunes compact LLMs for detecting cyberattacks on transformer current differential relays (TCDRs), which can otherwise cause false tripping of critical power transformers. The core idea is to textualize multivariate time-series current measurements from TCDRs, across phases and input/output sides, into structured natural-language prompts that are then processed by compact, locally deployable LLMs. Using this representation, we fine-tune DistilBERT, GPT-2, and DistilBERT+LoRA to distinguish cyberattacks from genuine fault-induced disturbances while preserving relay dependability. The proposed framework is evaluated against a broad set of state-of-the-art machine learning and deep learning baselines under nominal conditions, complex cyberattack scenarios, and measurement noise. Our results show that LLM-based detectors achieve competitive or superior cyberattack detection performance, with DistilBERT detecting up to 97.62% of attacks while maintaining perfect fault detection accuracy. Additional evaluations demonstrate robustness to prompt formulation variations, resilience under combined time-synchronization and false-data injection attacks, and stable performance under realistic measurement noise levels. The attention mechanisms of LLMs further enable intrinsic interpretability by highlighting the most influential time-phase regions of relay measurements. These results demonstrate that compact LLMs provide a practical, interpretable, and robust solution for enhancing cyberattack detection in modern digital substations. We provide the full dataset used in this study for reproducibility. --- ### Addressing Overthinking in Large Vision-Language Models via Gated Perception-Reasoning Optimization Authors: Xingjian Diao, Zheyuan Liu, Chunhui Zhang, Weiyi Wu, Keyi Kong, Lin Shi, Kaize Ding, Soroush Vosoughi, Jiang Gui Published: 2026-01-07 Categories: cs.CV, cs.CL arXiv: https://arxiv.org/abs/2601.04442v2 Large Vision-Language Models (LVLMs) have exhibited strong reasoning capabilities through chain-of-thought mechanisms that generate step-by-step rationales. However, such slow-thinking approaches often lead to overthinking, where models produce excessively verbose responses even for simple queries, resulting in test-time inefficiency and even degraded accuracy. Prior work has attempted to mitigate this issue via adaptive reasoning strategies, but these methods largely overlook a fundamental bottleneck: visual perception failures. We argue that stable reasoning critically depends on low-level visual grounding, and that reasoning errors often originate from imperfect perception rather than insufficient deliberation. To address this limitation, we propose Gated Perception-Reasoning Optimization (GPRO), a meta-reasoning controller that dynamically routes computation among three decision paths at each generation step: a lightweight fast path, a slow perception path for re-examining visual inputs, and a slow reasoning path for internal self-reflection. To learn this distinction, we derive large-scale failure attribution supervision from approximately 790k samples, using teacher models to distinguish perceptual hallucinations from reasoning errors. We then train the controller with multi-objective reinforcement learning to optimize the trade-off between task accuracy and computational cost under uncertainty. Experiments on five benchmarks demonstrate that GPRO substantially improves both accuracy and efficiency, outperforming recent slow-thinking methods while generating significantly shorter responses. --- ### Improving and Accelerating Offline RL in Large Discrete Action Spaces with Structured Policy Initialization Authors: Matthew Landers, Taylor W. Killian, Thomas Hartvigsen, Afsaneh Doryab Published: 2026-01-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.04441v2 Reinforcement learning in discrete combinatorial action spaces requires searching over exponentially many joint actions to simultaneously select multiple sub-actions that form coherent combinations. Existing approaches either simplify policy learning by assuming independence across sub-actions, which often yields incoherent or invalid actions, or attempt to learn action structure and control jointly, which is slow and unstable. We introduce Structured Policy Initialization (SPIN), a two-stage framework that first pre-trains an Action Structure Model (ASM) to capture the manifold of valid actions, then freezes this representation and trains lightweight policy heads for control. On challenging discrete DM Control benchmarks, SPIN improves average return by up to 39% over the state of the art while reducing time to convergence by up to 12.8$\times$. --- ### Learning to Simulate Human Dialogue Authors: Kanishk Gandhi, Agam Bhatia, Noah D. Goodman Published: 2026-01-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.04436v1 To predict what someone will say is to model how they think. We study this through next-turn dialogue prediction: given a conversation, predict the next utterance produced by a person. We compare learning approaches along two dimensions: (1) whether the model is allowed to think before responding, and (2) how learning is rewarded either through an LLM-as-a-judge that scores semantic similarity and information completeness relative to the ground-truth response, or by directly maximizing the log-probability of the true human dialogue. We find that optimizing for judge-based rewards indeed increases judge scores throughout training, however it decreases the likelihood assigned to ground truth human responses and decreases the win rate when human judges choose the most human-like response among a real and synthetic option. This failure is amplified when the model is allowed to think before answering. In contrast, by directly maximizing the log-probability of observed human responses, the model learns to better predict what people actually say, improving on both log-probability and win rate evaluations. Treating chain-of-thought as a latent variable, we derive a lower bound on the log-probability. Optimizing this objective yields the best results on all our evaluations. These results suggest that thinking helps primarily when trained with a distribution-matching objective grounded in real human dialogue, and that scaling this approach to broader conversational data may produce models with a more nuanced understanding of human behavior. --- ### Accommodation and Epistemic Vigilance: A Pragmatic Account of Why LLMs Fail to Challenge Harmful Beliefs Authors: Myra Cheng, Robert D. Hawkins, Dan Jurafsky Published: 2026-01-07 Categories: cs.CL, cs.AI, cs.CY arXiv: https://arxiv.org/abs/2601.04435v1 Large language models (LLMs) frequently fail to challenge users' harmful beliefs in domains ranging from medical advice to social reasoning. We argue that these failures can be understood and addressed pragmatically as consequences of LLMs defaulting to accommodating users' assumptions and exhibiting insufficient epistemic vigilance. We show that social and linguistic factors known to influence accommodation in humans (at-issueness, linguistic encoding, and source reliability) similarly affect accommodation in LLMs, explaining performance differences across three safety benchmarks that test models' ability to challenge harmful beliefs, spanning misinformation (Cancer-Myth, SAGE-Eval) and sycophancy (ELEPHANT). We further show that simple pragmatic interventions, such as adding the phrase "wait a minute", significantly improve performance on these benchmarks while preserving low false-positive rates. Our results highlight the importance of considering pragmatics for evaluating LLM behavior and improving LLM safety. --- ### CRUNet-MR-Univ: A Foundation Model for Diverse Cardiac MRI Reconstruction Authors: Donghang Lyu, Marius Staring, Hildo Lamb, Mariya Doneva Published: 2026-01-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.04428v1 In recent years, deep learning has attracted increasing attention in the field of Cardiac MRI (CMR) reconstruction due to its superior performance over traditional methods, particularly in handling higher acceleration factors, highlighting its potential for real-world clinical applications. However, current deep learning methods remain limited in generalizability. CMR scans exhibit wide variability in image contrast, sampling patterns, scanner vendors, anatomical structures, and disease types. Most existing models are designed to handle only a single or narrow subset of these variations, leading to performance degradation when faced with distribution shifts. Therefore, it is beneficial to develop a unified model capable of generalizing across diverse CMR scenarios. To this end, we propose CRUNet-MR-Univ, a foundation model that leverages spatio-temporal correlations and prompt-based priors to effectively handle the full diversity of CMR scans. Our approach consistently outperforms baseline methods across a wide range of settings, highlighting its effectiveness and promise. --- ### XGrammar-2: Dynamic and Efficient Structured Generation Engine for Agentic LLMs Authors: Linzhang Li, Yixin Dong, Guanjie Wang, Ziyi Xu, Alexander Jiang, Tianqi Chen Published: 2026-01-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.04426v4 Modern LLM agents increasingly rely on dynamic structured generation, such as tool calling and response protocols. Unlike traditional structured generation with static structures, these workloads vary both across requests and within a request, posing new challenges to existing engines. We present XGrammar-2, a structured generation engine for dynamic agentic workloads. Our design is based on two key ideas: first-class support for tag-triggered structure switching, and fine-grained reuse across requests with different output structures. Concretely, XGrammar-2 introduces TagDispatch for dynamic structural dispatching and Cross-Grammar Cache for substructure-level cache reuse across grammars. It further improves efficiency with an Earley-based adaptive token mask cache, just-in-time compilation, and repetition state compression. Experiments show that XGrammar-2 achieves over 6x faster compilation than prior structured generation engines, and incurs near-zero end-to-end overhead in modern LLM serving systems. --- ### Gavel: Agent Meets Checklist for Evaluating LLMs on Long-Context Legal Summarization Authors: Yao Dou, Benjamin Mamut, Wei Xu Published: 2026-01-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.04424v4 Large language models (LLMs) now support contexts of up to 1M tokens, but their strengths and weaknesses on complex long-context tasks remain unclear. To study this, we focus on multi-document legal case summarization, where a single case often spans many documents exceeding 100K tokens. We systematically evaluate 12 frontier LLMs with Gavel, which consists of Gavel-Ref, a reference-based evaluation framework with checklist, residual-fact, and writing-style evaluations, and Gavel-Agent, a reference-free agent for evaluating factual coverage directly from source documents. Our results show that current models are more prone to omitting key information than hallucinating. They all perform well on simple checklist items, such as filing date, but struggle with rare and complex items, such as settlements. Performance also declines as case length increases. To meta-evaluate Gavel, we collect 160 hours of human annotations. Gavel-Agent reduces token usage by at least 36% compared to end-to-end and chunk-by-chunk methods while achieving competitive performance. Gavel-Agent also generalizes to the medical domain, performing the best with at least 77% fewer tokens. --- ### Learning Multinomial Logits in $O(n \log n)$ time Authors: Flavio Chierichetti, Mirko Giacchini, Ravi Kumar, Silvio Lattanzi, Alessandro Panconesi, Erasmo Tani, Andrew Tomkins Published: 2026-01-07 Categories: cs.DS, cs.LG, math.ST, stat.ML arXiv: https://arxiv.org/abs/2601.04423v1 A Multinomial Logit (MNL) model is composed of a finite universe of items $(n)=\{1,..., n\}$, each assigned a positive weight. A query specifies an admissible subset -- called a slate -- and the model chooses one item from that slate with probability proportional to its weight. This query model is also known as the Plackett-Luce model or conditional sampling oracle in the literature. Although MNLs have been studied extensively, a basic computational question remains open: given query access to slates, how efficiently can we learn weights so that, for every slate, the induced choice distribution is within total variation distance $\varepsilon$ of the ground truth? This question is central to MNL learning and has direct implications for modern recommender system interfaces. We provide two algorithms for this task, one with adaptive queries and one with non-adaptive queries. Each algorithm outputs an MNL $M'$ that induces, for each slate $S$, a distribution $M'_S$ on $S$ that is within $\varepsilon$ total variation distance of the true distribution. Our adaptive algorithm makes $O\left(\frac{n}{\varepsilon^{3}}\log n\right)$ queries, while our non-adaptive algorithm makes $O\left(\frac{n^{2}}{\varepsilon^{3}}\log n \log\frac{n}{\varepsilon}\right)$ queries. Both algorithms query only slates of size two and run in time proportional to their query complexity. We complement these upper bounds with lower bounds of $Ω\left(\frac{n}{\varepsilon^{2}}\log n\right)$ for adaptive queries and $Ω\left(\frac{n^{2}}{\varepsilon^{2}}\log n\right)$ for non-adaptive queries, thus proving that our adaptive algorithm is optimal in its dependence on the support size $n$, while the non-adaptive one is tight within a $\log n$ factor. --- ### Transitive Expert Error and Routing Problems in Complex AI Systems Authors: Forest Mars Published: 2026-01-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.04416v1 Domain expertise enhances judgment within boundaries but creates systematic vulnerabilities specifically at borders. We term this Transitive Expert Error (TEE), distinct from Dunning-Kruger effects, requiring calibrated expertise as precondition. Mechanisms enabling reliable within-domain judgment become liabilities when structural similarity masks causal divergence. Two core mechanisms operate: structural similarity bias causes experts to overweight surface features (shared vocabulary, patterns, formal structure) while missing causal architecture differences; authority persistence maintains confidence across competence boundaries through social reinforcement and metacognitive failures (experts experience no subjective uncertainty as pattern recognition operates smoothly on familiar-seeming inputs.) These mechanism intensify under three conditions: shared vocabulary masking divergent processes, social pressure for immediate judgment, and delayed feedback. These findings extend to AI routing architectures (MoE systems, multi-model orchestration, tool-using agents, RAG systems) exhibiting routing-induced failures (wrong specialist selected) and coverage-induced failures (no appropriate specialist exists). Both produce a hallucination phenotype: confident, coherent, structurally plausible but causally incorrect outputs at domain boundaries. In human systems where mechanisms are cognitive black boxes; AI architectures make them explicit and addressable. We propose interventions: multi-expert activation with disagreement detection (router level), boundary-aware calibration (specialist level), and coverage gap detection (training level). TEE has detectable signatures (routing patterns, confidence-accuracy dissociations, domain-inappropriate content) enabling monitoring and mitigation. What remains intractable in human cognition becomes addressable through architectural design. --- ### Distribution-Guided and Constrained Quantum Machine Unlearning Authors: Nausherwan Malik, Zubair Khalid, Muhammad Faryad Published: 2026-01-07 Categories: cs.LG, quant-ph arXiv: https://arxiv.org/abs/2601.04413v2 Machine unlearning aims to remove the influence of specific training data from a learned model without full retraining. While recent work has begun to explore unlearning in quantum machine learning, existing approaches largely rely on fixed, uniform target distributions and do not explicitly control the trade-off between forgetting and retained model behaviour. In this work, we propose a distribution-guided framework for class-level quantum machine unlearning that treats unlearning as a constrained optimization problem. Our method introduces a tunable target distribution derived from model similarity statistics, decoupling the suppression of forgotten-class confidence from assumptions about redistribution among retained classes. We further incorporate an anchor-based preservation constraint that explicitly maintains predictive behaviour on selected retained data, yielding a controlled optimization trajectory that limits deviation from the original model. We evaluate the approach on variational quantum classifiers trained on the Iris and Covertype datasets. Results demonstrate sharp suppression of forgotten-class confidence, minimal degradation of retained-class performance, and closer alignment with the gold retrained model baselines compared to uniform-target unlearning. These findings highlight the importance of target design and constraint-based formulations for reliable and interpretable quantum machine unlearning. --- ### Rate or Fate? RLV$^\varepsilon$R: Reinforcement Learning with Verifiable Noisy Rewards Authors: Ali Rad, Khashayar Filom, Darioush Keivan, Peyman Mohajerin Esfahani, Ehsan Kamalinejad Published: 2026-01-07 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2601.04411v1 Reinforcement learning with verifiable rewards (RLVR) is a simple but powerful paradigm for training LLMs: sample a completion, verify it, and update. In practice, however, the verifier is almost never clean--unit tests probe only limited corner cases; human and synthetic labels are imperfect; and LLM judges (e.g., RLAIF) are noisy and can be exploited--and this problem worsens on harder domains (especially coding) where tests are sparse and increasingly model-generated. We ask a pragmatic question: does the verification noise merely slow down the learning (rate), or can it flip the outcome (fate)? To address this, we develop an analytically tractable multi-armed bandit view of RLVR dynamics, instantiated with GRPO and validated in controlled experiments. Modeling false positives and false negatives and grouping completions into recurring reasoning modes yields a replicator-style (natural-selection) flow on the probability simplex. The dynamics decouples into within-correct-mode competition and a one-dimensional evolution for the mass on incorrect modes, whose drift is determined solely by Youden's index J=TPR-FPR. This yields a sharp phase transition: when J>0, the incorrect mass is driven toward extinction (learning); when J=0, the process is neutral; and when J<0, incorrect modes amplify until they dominate (anti-learning and collapse). In the learning regime J>0, noise primarily rescales convergence time ("rate, not fate"). Experiments on verifiable programming tasks under synthetic noise reproduce the predicted J=0 boundary. Beyond noise, the framework offers a general lens for analyzing RLVR stability, convergence, and algorithmic interventions. --- ### From Preoperative CT to Postmastoidectomy Mesh Construction: Mastoidectomy Shape Prediction for Cochlear Implant Surgery Authors: Yike Zhang, Eduardo Davalos, Dingjie Su, Ange Lou, Jack Noble Published: 2026-01-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.04405v2 Cochlear Implant (CI) surgery treats severe hearing loss by inserting an electrode array into the cochlea to stimulate the auditory nerve. An important step in this procedure is mastoidectomy, which removes part of the mastoid region of the temporal bone to provide surgical access. Accurate mastoidectomy shape prediction from preoperative imaging improves pre-surgical planning, reduces risks, and enhances surgical outcomes. Despite its importance, there are limited deep-learning-based studies regarding this topic due to the challenges of acquiring ground-truth labels. We address this gap by investigating self-supervised and weakly-supervised learning models to predict the mastoidectomy region without human annotations. We propose a hybrid self-supervised and weakly-supervised learning framework to predict the mastoidectomy region directly from preoperative CT scans, where the mastoid remains intact. Our hybrid method achieves a mean Dice score of 0.72 when predicting the complex and boundary-less mastoidectomy shape, surpassing state-of-the-art approaches and demonstrating strong performance. The method provides groundwork for constructing 3D postmastoidectomy surfaces directly from the corresponding preoperative CT scans. To our knowledge, this is the first work that integrating self-supervised and weakly-supervised learning for mastoidectomy shape prediction, offering a robust and efficient solution for CI surgical planning while leveraging 3D T-distribution loss in weakly-supervised medical imaging. --- ### 3D-Agent:Tri-Modal Multi-Agent Collaboration for Scalable 3D Object Annotation Authors: Jusheng Zhang, Yijia Fan, Zimo Wen, Jian Wang, Keze Wang Published: 2026-01-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.04404v1 Driven by applications in autonomous driving robotics and augmented reality 3D object annotation presents challenges beyond 2D annotation including spatial complexity occlusion and viewpoint inconsistency Existing approaches based on single models often struggle to address these issues effectively We propose Tri MARF a novel framework that integrates tri modal inputs including 2D multi view images textual descriptions and 3D point clouds within a multi agent collaborative architecture to enhance large scale 3D annotation Tri MARF consists of three specialized agents a vision language model agent for generating multi view descriptions an information aggregation agent for selecting optimal descriptions and a gating agent that aligns textual semantics with 3D geometry for refined captioning Extensive experiments on Objaverse LVIS Objaverse XL and ABO demonstrate that Tri MARF substantially outperforms existing methods achieving a CLIPScore of 88 point 7 compared to prior state of the art methods retrieval accuracy of 45 point 2 and 43 point 8 on ViLT R at 5 and a throughput of up to 12000 objects per hour on a single NVIDIA A100 GPU --- ### Balancing Usability and Compliance in AI Smart Devices: A Privacy-by-Design Audit of Google Home, Alexa, and Siri Authors: Trevor De Clark, Yulia Bobkova, Ajay Kumar Shrestha Published: 2026-01-07 Categories: cs.CY, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2601.04403v2 This paper investigates the privacy and usability of AI-enabled smart devices commonly used by youth, focusing on Google Home Mini, Amazon Alexa, and Apple Siri. While these devices provide convenience and efficiency, they also raise privacy and transparency concerns due to their always-listening design and complex data management processes. The study proposes and applies a combined framework of Heuristic Evaluation, Personal Information Protection and Electronic Documents Act (PIPEDA) Compliance Assessment, and Youth-Centered Usability Testing to assess whether these devices align with Privacy-by-Design principles and support meaningful user control. Results show that Google Home achieved the highest usability score, while Siri scored highest in regulatory compliance, indicating a trade-off between user convenience and privacy protection. Alexa demonstrated clearer task navigation but weaker transparency in data retention. Findings suggest that although youth may feel capable of managing their data, their privacy self-efficacy remains limited by technical design, complex settings, and unclear data policies. The paper concludes that enhancing transparency, embedding privacy guidance during onboarding, and improving policy alignment are critical steps toward ensuring that smart devices are both usable and compliant with privacy standards that protect young users. --- ### Transformer-based Multi-agent Reinforcement Learning for Separation Assurance in Structured and Unstructured Airspaces Authors: Arsyi Aziz, Peng Wei Published: 2026-01-07 Categories: cs.RO, cs.LG, cs.MA arXiv: https://arxiv.org/abs/2601.04401v1 Conventional optimization-based metering depends on strict adherence to precomputed schedules, which limits the flexibility required for the stochastic operations of Advanced Air Mobility (AAM). In contrast, multi-agent reinforcement learning (MARL) offers a decentralized, adaptive framework that can better handle uncertainty, required for safe aircraft separation assurance. Despite this advantage, current MARL approaches often overfit to specific airspace structures, limiting their adaptability to new configurations. To improve generalization, we recast the MARL problem in a relative polar state space and train a transformer encoder model across diverse traffic patterns and intersection angles. The learned model provides speed advisories to resolve conflicts while maintaining aircraft near their desired cruising speeds. In our experiments, we evaluated encoder depths of 1, 2, and 3 layers in both structured and unstructured airspaces, and found that a single encoder configuration outperformed deeper variants, yielding near-zero near mid-air collision rates and shorter loss-of-separation infringements than the deeper configurations. Additionally, we showed that the same configuration outperforms a baseline model designed purely with attention. Together, our results suggest that the newly formulated state representation, novel design of neural network architecture, and proposed training strategy provide an adaptable and scalable decentralized solution for aircraft separation assurance in both structured and unstructured airspaces. --- ### Convenience vs. Control: A Qualitative Study of Youth Privacy with Smart Voice Assistants Authors: Molly Campbell, Trevor De Clark, Mohamad Sheikho Al Jasem, Sandhya Joshi, Ajay Kumar Shrestha Published: 2026-01-07 Categories: cs.CY, cs.AI, cs.HC arXiv: https://arxiv.org/abs/2601.04399v2 Smart voice assistants (SVAs) are embedded in the daily lives of youth, yet their privacy controls often remain opaque and difficult to manage. Through five semi-structured focus groups (N=26) with young Canadians (ages 16-24), we investigate how perceived privacy risks (PPR) and benefits (PPBf) intersect with algorithmic transparency and trust (ATT) and privacy self-efficacy (PSE) to shape privacy-protective behaviors (PPB). Our analysis reveals that policy overload, fragmented settings, and unclear data retention undermine self-efficacy and discourage protective actions. Conversely, simple transparency cues were associated with greater confidence without diminishing the utility of hands-free tasks and entertainment. We synthesize these findings into a qualitative model in which transparency friction erodes PSE, which in turn weakens PPB. From this model, we derive actionable design guidance for SVAs, including a unified privacy hub, plain-language "data nutrition" labels, clear retention defaults, and device-conditional micro-tutorials. This work foregrounds youth perspectives and offers a path for SVA governance and design that empowers young digital citizens while preserving convenience. --- ### Interpreting Transformers Through Attention Head Intervention Authors: Mason Kadem, Rong Zheng Published: 2026-01-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.04398v4 Neural networks are growing more capable on their own, but we do not understand their neural mechanisms. Understanding these mechanisms' decision-making processes, or mechanistic interpretability, enables (1) accountability and control in high-stakes domains, (2) the study of digital brains and the emergence of cognition, and (3) discovery of new knowledge when AI systems outperform humans. This paper traces how attention head intervention emerged as a key method for causal interpretability of transformers. The evolution from visualization to intervention represents a paradigm shift from observing correlations to causally validating mechanistic hypotheses through direct intervention. Head intervention studies revealed robust empirical findings while also highlighting limitations that complicate interpretation. Recent work demonstrates that mechanistic understanding now enables targeted control of model behaviour, successfully suppressing toxic outputs and manipulating semantic content through selective attention head intervention, validating the practical utility of interpretability research for AI safety. --- ### Performance Analysis of Image Classification on Bangladeshi Datasets Authors: Mohammed Sami Khan, Fabiha Muniat, Rowzatul Zannat Published: 2026-01-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.04397v1 Convolutional Neural Networks (CNNs) have demonstrated remarkable success in image classification tasks; however, the choice between designing a custom CNN from scratch and employing established pre-trained architectures remains an important practical consideration. In this work, we present a comparative analysis of a custom-designed CNN and several widely used deep learning architectures, including VGG-16, ResNet-50, and MobileNet, for an image classification task. The custom CNN is developed and trained from scratch, while the popular architectures are employed using transfer learning under identical experimental settings. All models are evaluated using standard performance metrics such as accuracy, precision, recall, and F1-score. Experimental results show that pre-trained CNN architectures consistently outperform the custom CNN in terms of classification accuracy and convergence speed, particularly when training data is limited. However, the custom CNN demonstrates competitive performance with significantly fewer parameters and reduced computational complexity. This study highlights the trade-offs between model complexity, performance, and computational efficiency, and provides practical insights into selecting appropriate CNN architectures for image classification problems. --- ### ARREST: Adversarial Resilient Regulation Enhancing Safety and Truth in Large Language Models Authors: Sharanya Dasgupta, Arkaprabha Basu, Sujoy Nath, Swagatam Das Published: 2026-01-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.04394v1 Human cognition, driven by complex neurochemical processes, oscillates between imagination and reality and learns to self-correct whenever such subtle drifts lead to hallucinations or unsafe associations. In recent years, LLMs have demonstrated remarkable performance in a wide range of tasks. However, they still lack human cognition to balance factuality and safety. Bearing the resemblance, we argue that both factual and safety failures in LLMs arise from a representational misalignment in their latent activation space, rather than addressing those as entirely separate alignment issues. We hypothesize that an external network, trained to understand the fluctuations, can selectively intervene in the model to regulate falsehood into truthfulness and unsafe output into safe output without fine-tuning the model parameters themselves. Reflecting the hypothesis, we propose ARREST (Adversarial Resilient Regulation Enhancing Safety and Truth), a unified framework that identifies and corrects drifted features, engaging both soft and hard refusals in addition to factual corrections. Our empirical results show that ARREST not only regulates misalignment but is also more versatile compared to the RLHF-aligned models in generating soft refusals due to adversarial training. We make our codebase available at https://github.com/sharanya-dasgupta001/ARREST. --- ### Assessing the quality and coherence of word embeddings after SCM-based intersectional bias mitigation Authors: Eren Kocadag, Seyed Sahand Mohammadi Ziabari, Ali Mohammed Mansoor Alsahag Published: 2026-01-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.04393v1 Static word embeddings often absorb social biases from the text they learn from, and those biases can quietly shape downstream systems. Prior work that uses the Stereotype Content Model (SCM) has focused mostly on single-group bias along warmth and competence. We broaden that lens to intersectional bias by building compound representations for pairs of social identities through summation or concatenation, and by applying three debiasing strategies: Subtraction, Linear Projection, and Partial Projection. We study three widely used embedding families (Word2Vec, GloVe, and ConceptNet Numberbatch) and assess them with two complementary views of utility: whether local neighborhoods remain coherent and whether analogy behavior is preserved. Across models, SCM-based mitigation carries over well to the intersectional case and largely keeps the overall semantic landscape intact. The main cost is a familiar trade off: methods that most tightly preserve geometry tend to be more cautious about analogy behavior, while more assertive projections can improve analogies at the expense of strict neighborhood stability. Partial Projection is reliably conservative and keeps representations steady; Linear Projection can be more assertive; Subtraction is a simple baseline that remains competitive. The choice between summation and concatenation depends on the embedding family and the application goal. Together, these findings suggest that intersectional debiasing with SCM is practical in static embeddings, and they offer guidance for selecting aggregation and debiasing settings when balancing stability against analogy performance. --- ### Enhanced-FQL($λ$), an Efficient and Interpretable RL with novel Fuzzy Eligibility Traces and Segmented Experience Replay Authors: Mohsen Jalaeian-Farimani, Xiong Xiong, Luca Bascetta Published: 2026-01-07 Categories: cs.LG, cs.AI, cs.RO, eess.SY, math.OC arXiv: https://arxiv.org/abs/2601.04392v2 This paper introduces a fuzzy reinforcement learning framework, Enhanced-FQL($λ$), that integrates novel Fuzzified Eligibility Traces (FET) and Segmented Experience Replay (SER) into fuzzy Q-learning with the Fuzzified Bellman Equation (FBE) for continuous control. The proposed approach employs an interpretable fuzzy rule base instead of complex neural architectures, while maintaining competitive performance through two key innovations: a fuzzified Bellman equation with eligibility traces for stable multi-step credit assignment, and a memory-efficient segment-based experience replay mechanism for enhanced sample efficiency. Theoretical analysis proves the proposed method convergence under standard assumptions. On the Cart--Pole benchmark, Enhanced-FQL($λ$) improves sample efficiency and reduces variance relative to $n$-step fuzzy TD and fuzzy SARSA($λ$), while remaining competitive with the tested DDPG baseline. These results support the proposed framework as an interpretable and computationally compact alternative for moderate-scale continuous control problems. --- ### SciFig: Towards Automating Editable Figure Generation for Scientific Papers Authors: Siyuan Huang, Yifan Zhou, Yutong Gao, Zi Yin, Juyang Bai, Xinxin Liu, Rama Chellappa, Chun Pong Lau, Cheng Peng, Sayan Nag, Shraman Pramanick Published: 2026-01-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.04390v3 High-quality methodology figures are central to scientific communication, yet they remain difficult and time-consuming to create. Such figures must distill a method's components and information flow into a clear, revisable diagram as the paper evolves. Existing methodology diagram automation systems typically face a trade-off between editability and visual quality: TikZ- or SVG-based methods produce editable structured outputs but often lack the richness of human-designed figures, while image-generation models produce polished raster outputs that are difficult to revise. We introduce SciFig, an end-to-end multi-agent framework for generating visually rich and fully editable methodology figures from scientific text. SciFig decomposes figure generation into planning, layout synthesis, component rendering, and iterative refinement, producing XML figures that can be edited in standard diagramming tools and refined through human or VLM feedback. We also introduce SciFig-Bench, a human-verified benchmark of 435 author-drawn methodology figures from 37 arXiv domains and 15 top-tier AI/ML venues, and SciFig-Eval, a four-axis evaluation protocol for measuring figure quality. Across seven single-agent and agentic baselines, SciFig achieves the best performance on all four SciFig-Eval axes and generates editable figures in about 10 minutes on average. Qualitative examples further show that SciFig can generalize beyond methodology figures to teaser diagrams and statistical plots. Dataset and code are available at: https://shramanpramanick.github.io/SciFig/. --- ### Safety Is Not Universal: The Selective Safety Trap in LLM Alignment Authors: Iago Alves Brito, Walcy Santos Rezende Rios, Julia Soares Dollis, Diogo Fernandes Costa Silva, Arlindo Rodrigues Galvão Filho Published: 2026-01-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2601.04389v3 Current safety evaluations of large language models (LLMs) create a dangerous illusion of universal protection by aggregating harms under generic categories such as "Identity Hate", obscuring vulnerabilities toward specific populations. In this work, we expose the Selective Safety Trap: a systemic failure mode where models robustly defend specific populations while leaving underrepresented communities highly vulnerable to identical adversarial attacks. To systematically audit this phenomenon, we introduce MiJaBench, a bilingual (English-Portuguese) adversarial benchmark comprising 43,961 controlled jailbreaking prompts across 16 minority groups. By evaluating 14 state-of-the-art LLMs on MiJaBench, we curate 615,454 prompt-response pairs that compose MiJaBench-Align, revealing that safety alignment is not a uniform semantic capability but a demographic hierarchy, with defense rates fluctuating by up to 42% within the same model solely based on the target group. This disparity persists across architectures and languages and is amplified by scaling, indicating that current alignment methods learn group-specific safeguards rather than a generalized notion of harm. Through targeted direct preference optimization (DPO) on a 1B-parameter baseline, we achieve strong zero-shot safety generalizations to entirely unseen demographics and complex attack strategies. We release all datasets and scripts to provide the community with a concrete pathway toward equitable, transferable safety alignment. --- ### LLM-Guided Lifecycle-Aware Clustering of Multi-Turn Customer Support Conversations Authors: Priyaranjan Pattnayak, Sanchari Chowdhuri, Amit Agarwal, Hitesh Laxmichand Patel Published: 2026-01-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2601.04388v1 Clustering customer chat data is vital for cloud providers handling multi service queries. Traditional methods struggle with overlapping concerns and create broad, static clusters that degrade over time. Reclustering disrupts continuity, making issue tracking difficult. We propose an adaptive system that segments multi turn chats into service specific concerns and incrementally refines clusters as new issues arise. Cluster quality is tracked via DaviesBouldin Index and Silhouette Scores, with LLM based splitting applied only to degraded clusters. Our method improves Silhouette Scores by over 100\% and reduces DBI by 65.6\% compared to baselines, enabling scalable, real time analytics without full reclustering. --- ### The Language of Bargaining: Linguistic Effects in LLM Negotiations Authors: Stuti Sinha, Himanshu Kumar, Aryan Raju Mandapati, Rakshit Sakhuja, Dhruv Kumar Published: 2026-01-07 Categories: cs.AI, cs.CL, cs.GT arXiv: https://arxiv.org/abs/2601.04387v2 Negotiation is a core component of social intelligence, requiring agents to balance strategic reasoning, cooperation, and social norms. Recent work shows that LLMs can engage in multi-turn negotiation, yet nearly all evaluations occur exclusively in English. Using controlled multi-agent simulations across Ultimatum, Buy-Sell, and Resource Exchange games, we systematically isolate language effects across English and four Indic framings (Hindi, Punjabi, Gujarati, Marwadi) by holding game rules, model parameters, and incentives constant across all conditions. We find that language choice can shift outcomes more strongly than changing models, reversing proposer advantages and reallocating surplus. Crucially, effects are task-contingent: Indic languages reduce stability in distributive games yet induce richer exploration in integrative settings. Our results demonstrate that evaluating LLM negotiation solely in English yields incomplete and potentially misleading conclusions. These findings caution against English-only evaluation of LLMs and suggest that culturally-aware evaluation is essential for fair deployment. --- ### Radiant Foam Rendering on a Graph Processor Authors: Zulkhuu Tuya, Ignacio Alzugaray, Nicholas Fry, Andrew J. Davison Published: 2026-01-07 Categories: cs.GR, cs.CV arXiv: https://arxiv.org/abs/2601.04382v2 Many emerging many-core accelerators replace a single large device memory with hundreds to thousands of lightweight cores, each owning only a small local SRAM and exchanging data via explicit on-chip communication. This organization offers high aggregate bandwidth, but it breaks a key assumption behind many volumetric rendering techniques: that rays can randomly access a large, unified scene representation. Rendering efficiently on such hardware therefore requires distributing both data and computation, keeping ray traversal mostly local, and structuring communication into predictable routes. We present a fully in-SRAM, distributed renderer for the Radiant Foam Voronoi-cell volumetric representation on the Graphcore Mk2 IPU(Intelligence Processing Unit), a many-core accelerator with tile-local SRAM and explicit inter-tile communication. Our system shards the scene across tiles and forwards rays between shards through a hierarchical routing overlay, enabling ray marching entirely from on-chip SRAM with predictable communication. On Mip-NeRF~360 scenes, the system attains near-interactive throughput of approximately 1 fps at 640x480 with image and depth map quality close to the original GPU-based Radiant Foam implementation, while keeping all scene data and ray state in on-chip SRAM. Beyond demonstrating feasibility, we analyze routing, memory, and scheduling bottlenecks that inform how future distributed-memory accelerators can better support irregular, data-movement-heavy rendering workloads. --- ### Few-Shot LoRA Adaptation of a Flow-Matching Foundation Model for Cross-Spectral Object Detection Authors: Maxim Clouser, Kia Khezeli, John Kalantari Published: 2026-01-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.04381v1 Foundation models for vision are predominantly trained on RGB data, while many safety-critical applications rely on non-visible modalities such as infrared (IR) and synthetic aperture radar (SAR). We study whether a single flow-matching foundation model pre-trained primarily on RGB images can be repurposed as a cross-spectral translator using only a few co-measured examples, and whether the resulting synthetic data can enhance downstream detection. Starting from FLUX.1 Kontext, we insert low-rank adaptation (LoRA) modules and fine-tune them on just 100 paired images per domain for two settings: RGB to IR on the KAIST dataset and RGB to SAR on the M4-SAR dataset. The adapted model translates RGB images into pixel-aligned IR/SAR, enabling us to reuse existing bounding boxes and train object detection models purely in the target modality. Across a grid of LoRA hyperparameters, we find that LPIPS computed on only 50 held-out pairs is a strong proxy for downstream performance: lower LPIPS consistently predicts higher mAP for YOLOv11n on both IR and SAR, and for DETR on KAIST IR test data. Using the best LPIPS-selected LoRA adapter, synthetic IR from external RGB datasets (LLVIP, FLIR ADAS) improves KAIST IR pedestrian detection, and synthetic SAR significantly boosts infrastructure detection on M4-SAR when combined with limited real SAR. Our results suggest that few-shot LoRA adaptation of flow-matching foundation models is a promising path toward foundation-style support for non-visible modalities. --- ### Aligned explanations in neural networks Authors: Corentin Lobet, Francesca Chiaromonte Published: 2026-01-07 Categories: cs.LG, cs.CV, stat.ML arXiv: https://arxiv.org/abs/2601.04378v3 As artificial intelligence increasingly drives critical decisions, the ability to genuinely explain how neural networks make predictions is essential for trust. Yet, most current explanation methods offer post-hoc rationalizations rather than guaranteeing a true reflection of the model's reasoning. We introduce the notion of explanatory alignment, a requirement that explanations directly construct predictions rather than rationalize them. To achieve this in complex data domains, we present Pointwise-interpretable Networks (PiNets), a pseudo-linear architecture that forms linear models instance-wise. Evaluated on image classification and segmentation tasks, PiNets demonstrate that their explanations are deeply faithful across four criteria: meaningfulness, alignment, robustness, and sufficiency (MARS). Our contributions pave the way for promising avenues: by reconciling the predictive power of deep learning with the interpretability of linear models, PiNets provide a principled foundation for trustworthy AI and data-driven scientific discovery. --- ### Disco-RAG: Discourse-Aware Retrieval-Augmented Generation Authors: Dongqi Liu, Hang Ding, Qiming Feng, Xurong Xie, Zhucun Xue, Chengjie Wang, Jian Li, Jiangning Zhang, Yabiao Wang Published: 2026-01-07 Categories: cs.CL, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2601.04377v5 Retrieval-Augmented Generation (RAG) has emerged as an important means of enhancing the performance of large language models (LLMs) in knowledge-intensive tasks. However, most existing RAG strategies treat retrieved passages in a flat and unstructured way, which prevents the model from capturing structural cues and constrains its ability to synthesize knowledge from dispersed evidence across documents. To overcome these limitations, we propose Disco-RAG, a discourse-aware framework that explicitly injects discourse signals into the generation process. Our method constructs intra-chunk discourse trees to capture local hierarchies and builds inter-chunk rhetorical graphs to model cross-passage coherence. These structures are jointly integrated into a planning blueprint that conditions the generation. Experiments on question answering and long-document summarization benchmarks show the efficacy of our approach. Disco-RAG achieves state-of-the-art results on the benchmarks without fine-tuning. These findings underscore the important role of discourse structure in advancing RAG systems. --- ### Combining Facial Videos and Biosignals for Stress Estimation During Driving Authors: Paraskevi Valergaki, Vassilis C. Nicodemou, Iason Oikonomidis, Antonis Argyros, Anastasios Roussos Published: 2026-01-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.04376v3 Reliable stress recognition is critical in applications such as medical monitoring and safety-critical systems, including real-world driving. While stress is commonly detected using physiological signals such as perinasal perspiration and heart rate, facial activity provides complementary cues that can be captured unobtrusively from video. We propose a multimodal stress estimation framework that combines facial videos and physiological signals, remaining effective even when biosignal acquisition is challenging. Facial behavior is represented using a dense 3D Morphable Model, yielding a 56-dimensional descriptor that captures subtle expression and head-pose dynamics over time. To study how stress modulates facial motion, we perform extensive experiments alongside established physiological markers. Paired hypothesis tests between baseline and stressor phases show that 38 of 56 facial components exhibit consistent, phase-specific stress responses comparable to physiological markers. Building on these findings, we introduce a Transformer-based temporal modeling framework and evaluate unimodal, early-fusion, and cross-modal attention strategies. Cross-modal attention fusion of 3D-derived facial features with physiological signals substantially improves performance over physiological signals alone, increasing AUROC from 52.7% and accuracy from 51.0% to 92.0% and 86.7%, respectively. Although evaluated on driving data, the proposed framework and protocol may generalize to other stress estimation settings. --- ### Dialect Matters: Cross-Lingual ASR Transfer for Low-Resource Indic Language Varieties Authors: Akriti Dhasmana, Aarohi Srivastava, David Chiang Published: 2026-01-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.04373v2 We conduct an empirical study of cross-lingual transfer using spontaneous, noisy, and code-mixed speech across a wide range of Indic dialects and language varieties. Our results indicate that although ASR performance is generally improved with reduced phylogenetic distance between languages, this factor alone does not fully explain performance in dialectal settings. Often, fine-tuning on smaller amounts of dialectal data yields performance comparable to fine-tuning on larger amounts of phylogenetically-related, high-resource standardized languages. We also present a case study on Garhwali, a low-resource Pahari language variety, and evaluate multiple contemporary ASR models. Finally, we analyze transcription errors to examine bias toward pre-training languages, providing additional insight into challenges faced by ASR systems on dialectal and non-standardized speech. --- ### End-to-end differentiable design of geometric waveguide displays Authors: Xinge Yang, Zhaocheng Liu, Zhaoyu Nie, Qingyuan Fan, Zhimin Shi, Jim Bonar, Wolfgang Heidrich Published: 2026-01-07 Categories: physics.optics, cs.CV, cs.GR arXiv: https://arxiv.org/abs/2601.04370v1 Geometric waveguides are a promising architecture for optical see-through augmented reality displays, but their performance is severely bottlenecked by the difficulty of jointly optimizing non-sequential light transport and polarization-dependent multilayer thin-film coatings. Here we present the first end-to-end differentiable optimization framework for geometric waveguide that couples non-sequential Monte Carlo polarization ray tracing with a differentiable transfer-matrix thin-film solver. A differentiable Monte Carlo ray tracer avoids the exponential growth of deterministic ray splitting while enabling gradients backpropagation from eyebox metrics to design parameters. With memory-saving strategies, we optimize more than one thousand layer-thickness parameters and billions of non-sequential ray-surface intersections on a single multi-GPU workstation. Automated layer pruning is achieved by starting from over-parameterized stacks and driving redundant layers to zero thickness under discrete manufacturability constraints, effectively performing topology optimization to discover optimal coating structures. On a representative design, starting from random initialization within thickness bounds, our method increases light efficiency from 4.1\% to 33.5\% and improves eyebox and FoV uniformity by $\sim$17$\times$ and $\sim$11$\times$, respectively. Furthermore, we jointly optimize the waveguide and an image preprocessing network to improve perceived image quality. Our framework not only enables system-level, high-dimensional coating optimization inside the waveguide, but also expands the scope of differentiable optics for next-generation optical design. --- ### Generalization to Political Beliefs from Fine-Tuning on Sports Team Preferences Authors: Owen Terry Published: 2026-01-07 Categories: physics.soc-ph, cs.CL arXiv: https://arxiv.org/abs/2601.04369v3 Fine-tuned LLMs often exhibit unexpected behavior as a result of generalizing beyond the data they're shown. We present results in which an LLM fine-tuned to prefer either coastal sports teams or Southern sports teams adopt political beliefs that diverge significantly from those of the base model. While we hypothesized that the coastal model would become more liberal and the southern model would become more conservative, we find that their responses are usually similar to each other, without a clear-cut liberal or conservative bias. In addition to asking the models for numerical ratings of agreement with relevant political statements, we ask them to elaborate on their more radical answers, finding varying degrees of willingness to justify themselves. Further work is needed to understand the mechanisms by which fine-tuning on simple, narrow datasets leads to seemingly unrelated changes in model behavior. --- ### Graph Integrated Transformers for Community Detection in Social Networks Authors: Heba Zahran, M. Omair Shafiq Published: 2026-01-07 Categories: cs.SI, cs.AI arXiv: https://arxiv.org/abs/2601.04367v1 Community detection is crucial for applications like targeted marketing and recommendation systems. Traditional methods rely on network structure, and embedding-based models integrate semantic information. However, there is a challenge when a model leverages local and global information from complex structures like social networks. Graph Neural Networks (GNNs) and Transformers have shown superior performance in capturing local and global relationships. In this paper, We propose Graph Integrated Transformer for Community Detection (GIT-CD), a hybrid model combining GNNs and Transformer-based attention mechanisms to enhance community detection in social networks. Specifically, the GNN module captures local graph structures, while the Transformer module models long-range dependencies. A self-optimizing clustering module refines community assignments using K-Means, silhouette loss, and KL divergence minimization. Experimental results on benchmark datasets show that GIT-CD outperforms state-of-the-art models, making it a robust approach for detecting meaningful communities in complex social networks. --- ### Machine Learning Model for Sparse PCM Completion Authors: Selcuk Koyuncu, Ronak Nouri, Stephen Providence Published: 2026-01-07 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2601.04366v1 In this paper, we propose a machine learning model for sparse pairwise comparison matrices (PCMs), combining classical PCM approaches with graph-based learning techniques. Numerical results are provided to demonstrate the effectiveness and scalability of the proposed method. --- ### Survival Dynamics of Neural and Programmatic Policies in Evolutionary Reinforcement Learning Authors: Anton Roupassov-Ruiz, Yiyang Zuo Published: 2026-01-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2601.04365v2 In evolutionary reinforcement learning tasks (ERL), agent policies are often encoded as small artificial neural networks (NERL). Such representations lack explicit modular structure, limiting behavioral interpretation. We investigate whether programmatic policies (PERL), implemented as soft, differentiable decision lists (SDDL), can match the performance of NERL. To support reproducible evaluation, we provide the first fully specified and open-source reimplementation of the classic 1992 Artificial Life (ALife) ERL testbed. We conduct a rigorous survival analysis across 4000 independent trials utilizing Kaplan-Meier curves and Restricted Mean Survival Time (RMST) metrics absent in the original study. We find a statistically significant difference in survival probability between PERL and NERL. PERL agents survive on average 201.69 steps longer than NERL agents. Moreover, SDDL agents using learning alone (no evolution) survive on average 73.67 steps longer than neural agents using both learning and evaluation. These results demonstrate that programmatic policies can exceed the survival performance of neural policies in ALife. --- ### Phasor Agents: Oscillatory Graphs with Three-Factor Plasticity and Sleep-Staged Learning Authors: Rodja Trappe Published: 2026-01-07 Categories: cs.LG, cs.NE, q-bio.NC arXiv: https://arxiv.org/abs/2601.04362v1 Phasor Agents are dynamical systems whose internal state is a Phasor Graph: a weighted graph of coupled Stuart-Landau oscillators. A Stuart-Landau oscillator is a minimal stable "rhythm generator" (the normal form near a Hopf bifurcation); each oscillator is treated as an abstract computational unit (inspired by, but not claiming to model, biological oscillatory populations). In this interpretation, oscillator phase tracks relative timing (coherence), while amplitude tracks local gain or activity. Relative phase structure serves as a representational medium; coupling weights are learned via three-factor local plasticity - eligibility traces gated by sparse global modulators and oscillation-timed write windows - without backpropagation. A central challenge in oscillatory substrates is stability: online weight updates can drive the network into unwanted regimes (e.g., global synchrony), collapsing representational diversity. We therefore separate wake tagging from offline consolidation, inspired by synaptic tagging-and-capture and sleep-stage dynamics: deep-sleep-like gated capture commits tagged changes safely, while REM-like replay reconstructs and perturbs experience for planning. A staged experiment suite validates each mechanism with ablations and falsifiers: eligibility traces preserve credit under delayed modulation; compression-progress signals pass timestamp-shuffle controls; phase-coherent retrieval reaches 4x diffusive baselines under noise; wake/sleep separation expands stable learning by 67 percent under matched weight-norm budgets; REM replay improves maze success rate by +45.5 percentage points; and a Tolman-style latent-learning signature - immediate competence and detour advantage after unrewarded exploration, consistent with an internal model - emerges from replay (Tolman, 1948). The codebase and all artifacts are open-source. --- ### Causally-Aware Information Bottleneck for Domain Adaptation Authors: Mohammad Ali Javidian Published: 2026-01-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2601.04361v1 We tackle a common domain adaptation setting in causal systems. In this setting, the target variable is observed in the source domain but is entirely missing in the target domain. We aim to impute the target variable in the target domain from the remaining observed variables under various shifts. We frame this as learning a compact, mechanism-stable representation. This representation preserves information relevant for predicting the target while discarding spurious variation. For linear Gaussian causal models, we derive a closed-form Gaussian Information Bottleneck (GIB) solution. This solution reduces to a canonical correlation analysis (CCA)-style projection and offers Directed Acyclic Graph (DAG)-aware options when desired. For nonlinear or non-Gaussian data, we introduce a Variational Information Bottleneck (VIB) encoder-predictor. This approach scales to high dimensions and can be trained on source data and deployed zero-shot to the target domain. Across synthetic and real datasets, our approach consistently attains accurate imputations, supporting practical use in high-dimensional causal models and furnishing a unified, lightweight toolkit for causal domain adaptation. --- ### PackCache: A Training-Free Acceleration Method for Unified Autoregressive Video Generation via Compact KV-Cache Authors: Kunyang Li, Mubarak Shah, Yuzhang Shang Published: 2026-01-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.04359v1 A unified autoregressive model is a Transformer-based framework that addresses diverse multimodal tasks (e.g., text, image, video) as a single sequence modeling problem under a shared token space. Such models rely on the KV-cache mechanism to reduce attention computation from O(T^2) to O(T); however, KV-cache size grows linearly with the number of generated tokens, and it rapidly becomes the dominant bottleneck limiting inference efficiency and generative length. Unified autoregressive video generation inherits this limitation. Our analysis reveals that KV-cache tokens exhibit distinct spatiotemporal properties: (i) text and conditioning-image tokens act as persistent semantic anchors that consistently receive high attention, and (ii) attention to previous frames naturally decays with temporal distance. Leveraging these observations, we introduce PackCache, a training-free KV-cache management method that dynamically compacts the KV cache through three coordinated mechanisms: condition anchoring that preserves semantic references, cross-frame decay modeling that allocates cache budget according to temporal distance, and spatially preserving position embedding that maintains coherent 3D structure under cache removal. In terms of efficiency, PackCache accelerates end-to-end generation by 1.7-2.2x on 48-frame long sequences, showcasing its strong potential for enabling longer-sequence video generation. Notably, the final four frames - the portion most impacted by the progressively expanding KV-cache and thus the most expensive segment of the clip - PackCache delivers a 2.6x and 3.7x acceleration on A40 and H200, respectively, for 48-frame videos. --- ### UNIC: Learning Unified Multimodal Extrinsic Contact Estimation Authors: Zhengtong Xu, Yuki Shirai Published: 2026-01-07 Categories: cs.RO, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2601.04356v2 Contact-rich manipulation requires reliable estimation of extrinsic contacts-the interactions between a grasped object and its environment which provide essential contextual information for planning, control, and policy learning. However, existing approaches often rely on restrictive assumptions, such as predefined contact types, fixed grasp configurations, or camera calibration, that hinder generalization to novel objects and deployment in unstructured environments. In this paper, we present UNIC, a unified multimodal framework for extrinsic contact estimation that operates without any prior knowledge or camera calibration. UNIC directly encodes visual observations in the camera frame and integrates them with proprioceptive and tactile modalities in a fully data-driven manner. It introduces a unified contact representation based on scene affordance maps that captures diverse contact formations and employs a multimodal fusion mechanism with random masking, enabling robust multimodal representation learning. Extensive experiments demonstrate that UNIC performs reliably. It achieves a 9.6 mm average Chamfer distance error on unseen contact locations, performs well on unseen objects, remains robust under missing modalities, and adapts to dynamic camera viewpoints. These results establish extrinsic contact estimation as a practical and versatile capability for contact-rich manipulation. The overview and hardware experiment videos are at https://youtu.be/xpMitkxN6Ls?si=7Vgj-aZ_P1wtnWZN --- ### Comparative Analysis of Custom CNN Architectures versus Pre-trained Models and Transfer Learning: A Study on Five Bangladesh Datasets Authors: Ibrahim Tanvir, Alif Ruslan, Sartaj Solaiman Published: 2026-01-07 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2601.04352v1 This study presents a comprehensive comparative analysis of custom-built Convolutional Neural Networks (CNNs) against popular pre-trained architectures (ResNet-18 and VGG-16) using both feature extraction and transfer learning approaches. We evaluated these models across five diverse image classification datasets from Bangladesh: Footpath Vision, Auto Rickshaw Detection, Mango Image Classification, Paddy Variety Recognition, and Road Damage Detection. Our experimental results demonstrate that transfer learning with fine-tuning consistently outperforms both custom CNNs built from scratch and feature extraction methods, achieving accuracy improvements ranging from 3% to 76% across different datasets. Notably, ResNet-18 with fine-tuning achieved perfect 100% accuracy on the Road Damage BD dataset. While custom CNNs offer advantages in model size (3.4M parameters vs. 11-134M for pre-trained models) and training efficiency on simpler tasks, pre-trained models with transfer learning provide superior performance, particularly on complex classification tasks with limited training data. This research provides practical insights for practitioners in selecting appropriate deep learning approaches based on dataset characteristics, computational resources, and performance requirements. --- ### RIGOURATE: Quantifying Scientific Exaggeration with Evidence-Aligned Claim Evaluation Authors: Joseph James, Chenghao Xiao, Yucheng Li, Nafise Sadat Moosavi, Chenghua Lin Published: 2026-01-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2601.04350v2 Scientific rigour tends to be sidelined in favour of bold statements, leading authors to overstate claims beyond what their results support. We present RIGOURATE, a two-stage multimodal framework that retrieves supporting evidence from a paper's body and assigns each claim an overstatement score. The framework consists of a dataset of over 10K claim-evidence sets from ICLR and NeurIPS papers, annotated using eight LLMs, with overstatement scores calibrated using peer-review comments and validated through human evaluation. It employes a fine-tuned reranker for evidence retrieval and a fine-tuned model to predict overstatement scores with justification. Compared to strong baselines, RIGOURATE enables improved evidence retrieval and overstatement detection. Overall, our work operationalises evidential proportionality and supports clearer, more transparent scientific communication. --- ### SCAR-GS: Spatial Context Attention for Residuals in Progressive Gaussian Splatting Authors: Diego Revilla, Pooja Suresh, Ooi Wei Tsang, Anand Bhojan Published: 2026-01-07 Categories: cs.CV, cs.GR arXiv: https://arxiv.org/abs/2601.04348v2 Recent advances in 3D Gaussian Splatting have allowed for real-time, high-fidelity novel view synthesis. Nonetheless, these models have significant storage requirements for large and medium-sized scenes, hindering their deployment over cloud and streaming services. Some of the most recent progressive compression techniques for these models rely on progressive masking and scalar quantization techniques to reduce the bitrate of Gaussian attributes using spatial context models. While effective, scalar quantization may not optimally capture the correlations of high-dimensional feature vectors, which can potentially limit the rate-distortion performance. In this work, we introduce a novel progressive codec for 3D Gaussian Splatting that replaces traditional methods with a more powerful Residual Vector Quantization approach to compress the primitive features. Our key contribution is an auto-regressive entropy model, guided by a multi-resolution hash grid, that accurately predicts the conditional probability of each successive transmitted index, allowing for coarse and refinement layers to be compressed with high efficiency. --- ### Summary of The Inaugural Music Source Restoration Challenge Authors: Yongyi Zang, Jiarui Hai, Wanying Ge, Qiuqiang Kong, Zheqi Dai, Helin Wang, Yuki Mitsufuji, Mark D. Plumbley Published: 2026-01-07 Categories: cs.SD, cs.AI, eess.AS arXiv: https://arxiv.org/abs/2601.04343v1 Music Source Restoration (MSR) aims to recover original, unprocessed instrument stems from professionally mixed and degraded audio, requiring the reversal of both production effects and real-world degradations. We present the inaugural MSR Challenge, which features objective evaluation on studio-produced mixtures using Multi-Mel-SNR, Zimtohrli, and FAD-CLAP, alongside subjective evaluation on real-world degraded recordings. Five teams participated in the challenge. The winning system achieved 4.46 dB Multi-Mel-SNR and 3.47 MOS-Overall, corresponding to relative improvements of 91% and 18% over the second-place system, respectively. Per-stem analysis reveals substantial variation in restoration difficulty across instruments, with bass averaging 4.59 dB across all teams, while percussion averages only 0.29 dB. The dataset, evaluation protocols, and baselines are available at https://msrchallenge.com/. --- ### ReHyAt: Recurrent Hybrid Attention for Video Diffusion Transformers Authors: Mohsen Ghafoorian, Amirhossein Habibian Published: 2026-01-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2601.04342v1 Recent advances in video diffusion models have shifted towards transformer-based architectures, achieving state-of-the-art video generation but at the cost of quadratic attention complexity, which severely limits scalability for longer sequences. We introduce ReHyAt, a Recurrent Hybrid Attention mechanism that combines the fidelity of softmax attention with the efficiency of linear attention, enabling chunk-wise recurrent reformulation and constant memory usage. Unlike the concurrent linear-only SANA Video, ReHyAt's hybrid design allows efficient distillation from existing softmax-based models, reducing the training cost by two orders of magnitude to ~160 GPU hours, while being competitive in the quality. Our light-weight distillation and finetuning pipeline provides a recipe that can be applied to future state-of-the-art bidirectional softmax-based models. Experiments on VBench and VBench-2.0, as well as a human preference study, demonstrate that ReHyAt achieves state-of-the-art video quality while reducing attention cost from quadratic to linear, unlocking practical scalability for long-duration and on-device video generation. Project page is available at https://qualcomm-ai-research.github.io/rehyat. --- ### Unified Text-Image Generation with Weakness-Targeted Post-Training Authors: Jiahui Chen, Philippe Hansen-Estruch, Xiaochuang Han, Yushi Hu, Emily Dinan, Amita Kamath, Michal Drozdzal, Reyhane Askari-Hemmat, Luke Zettlemoyer, Marjan Ghazvininejad Published: 2026-01-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2601.04339v3 Unified multimodal generation architectures that jointly produce text and images have recently emerged as a promising direction for text-to-image (T2I) synthesis. However, many existing systems rely on explicit modality switching, generating reasoning text before switching manually to image generation. This separate, sequential inference process limits cross-modal coupling and prohibits automatic multimodal generation. This work explores post-training to achieve fully unified text-image generation, where a model autonomously transitions from textual reasoning to visual synthesis within a single inference process. We study this on BAGEL, a 14B mixture-of-transformers model that pairs autoregressive text generation with flow-matching image synthesis. We examine the impact of joint text-image generation on T2I performance and the relative importance of each modality during post-training. We additionally explore different post-training data strategies, showing that a targeted dataset addressing specific limitations achieves superior results compared to broad image-caption corpora or benchmark-aligned data. Using offline, reward-weighted post-training with fully self-generated synthetic data, our approach enables improvements in multimodal image generation across four diverse, independent T2I benchmarks, demonstrating the effectiveness of reward-weighting both modalities and strategically designed post-training data. ---