Authors: Sahil Garg, Brad Cheezum, Sridhar Dutta, Vishal Agarwal
Abstract: Judge agents are fundamental to agentic AI frameworks: they provide automated evaluation, and enable iterative self-refinement of reasoning processes. We introduce JAF: Judge Agent Forest, a framework in which the judge agent conducts joint inference across a cohort of query--response pairs generated by a primary agent, rather than evaluating each in isolation. This paradigm elevates the judge from a local evaluator to a holistic learner: by simultaneously assessing related responses, the judge discerns cross-instance patterns and inconsistencies, whose aggregate feedback enables the primary agent to improve by viewing its own outputs through the judge's collective perspective. Conceptually, JAF bridges belief propagation and ensemble-learning principles: overlapping in-context neighborhoods induce a knowledge-graph structure that facilitates propagation of critique, and repeated, randomized evaluations yield a robust ensemble of context-sensitive judgments. JAF can be instantiated entirely via ICL, with the judge prompted for each query using its associated primary-agent response plus a small, possibly noisy set of peer exemplars. While kNN in embedding space is a natural starting point for exemplars, this approach overlooks categorical structure, domain metadata, or nuanced distinctions accessible to modern LLMs. To overcome these limitations, we develop a flexible locality-sensitive hashing (LSH) algorithm that learns informative binary codes by integrating semantic embeddings, LLM-driven hash predicates, supervision from categorical labels, and relevant side information. These hash codes support efficient, interpretable, and relation-aware selection of diverse exemplars, and further optimize exploration of CoT reasoning paths. We validate JAF with an empirical study on the demanding task of cloud misconfigs triage in large-scale cloud environments.
Authors: Khush Patel, Siva Surendira, Jithin George, Shreyas Kapale
Abstract: Large Language Models demonstrate remarkable capabilities yet remain fundamentally probabilistic, presenting critical reliability challenges for enterprise deployment. We introduce the Six Sigma Agent, a novel architecture that achieves enterprise-grade reliability through three synergistic components: (1) task decomposition into a dependency tree of atomic actions; (2) micro-agent sampling where each task is executed n times in parallel across diverse LLMs to generate independent outputs; and (3) consensus voting with dynamic scaling, clustering outputs and selecting the answer from the winning cluster with maximum votes. We prove that sampling n independent outputs with error rate p achieves system error O(p^{ceil(n/2)}), enabling exponential reliability gains. Even using cheaper models with 5% per-action error, consensus voting with 5 agents reduces error to 0.11%; dynamic scaling to 13 agents achieves 3.4 DPMO (Defects Per Million Opportunities), the Six Sigma standard. Evaluation across three enterprise use cases demonstrates a 14,700x reliability improvement over single-agent execution while reducing costs by 80%. Our work establishes that reliability in AI systems emerges from principled redundancy and consensus rather than model scaling alone.
Authors: Zehong Wang, Fang Wu, Hongru Wang, Xiangru Tang, Bolian Li, Zhenfei Yin, Yijun Ma, Yiyang Li, Weixiang Sun, Xiusi Chen, Yanfang Ye
Abstract: Large language model (LLM)-based agents exhibit strong step-by-step reasoning capabilities over short horizons, yet often fail to sustain coherent behavior over long planning horizons. We argue that this failure reflects a fundamental mismatch: step-wise reasoning induces a form of step-wise greedy policy that is adequate for short horizons but fails in long-horizon planning, where early actions must account for delayed consequences. From this planning-centric perspective, we study LLM-based agents in deterministic, fully structured environments with explicit state transitions and evaluation signals. Our analysis reveals a core failure mode of reasoning-based policies: locally optimal choices induced by step-wise scoring lead to early myopic commitments that are systematically amplified over time and difficult to recover from. We introduce FLARE (Future-aware Lookahead with Reward Estimation) as a minimal instantiation of future-aware planning to enforce explicit lookahead, value propagation, and limited commitment in a single model, allowing downstream outcomes to influence early decisions. Across multiple benchmarks, agent frameworks, and LLM backbones, FLARE consistently improves task performance and planning-level behavior, frequently allowing LLaMA-8B with FLARE to outperform GPT-4o with standard step-by-step reasoning. These results establish a clear distinction between reasoning and planning.
Authors: Ala N. Tak, Amin Banayeeanzade, Anahita Bolourani, Fatemeh Bahrani, Ashutosh Chaubey, Sai Praneeth Karimireddy, Norbert Schwarz, Jonathan Gratch
Abstract: Large Language Models (LLMs) are increasingly positioned as decision engines for hiring, healthcare, and economic judgment, yet real-world human judgment reflects a balance between rational deliberation and emotion-driven bias. If LLMs are to participate in high-stakes decisions or serve as models of human behavior, it is critical to assess whether they exhibit analogous patterns of (ir)rationalities and biases. To this end, we evaluate multiple LLM families on (i) benchmarks testing core axioms of rational choice and (ii) classic decision domains from behavioral economics and social norms where emotions are known to shape judgment and choice. Across settings, we show that deliberate "thinking" reliably improves rationality and pushes models toward expected-value maximization. To probe human-like affective distortions and their interaction with reasoning, we use two emotion-steering methods: in-context priming (ICP) and representation-level steering (RLS). ICP induces strong directional shifts that are often extreme and difficult to calibrate, whereas RLS produces more psychologically plausible patterns but with lower reliability. Our results suggest that the same mechanisms that improve rationality also amplify sensitivity to affective interventions, and that different steering methods trade off controllability against human-aligned behavior. Overall, this points to a tension between reasoning and affective steering, with implications for both human simulation and the safe deployment of LLM-based decision systems.
Authors: Yujie Hui, Xiaoyi Lu, Andrew Perrault, Yang Wang
Abstract: Provably correct distributed protocols, which are a critical component of modern distributed systems, are highly challenging to design and have often required decades of human effort. These protocols allow multiple agents to coordinate to come to a common agreement in an environment with uncertainty and failures. We formulate protocol design as a search problem over strategies in a game with imperfect information, and the desired correctness conditions are specified in Satisfiability Modulo Theories (SMT). However, standard methods for solving multi-agent games fail to learn correct protocols in this setting, even when the number of agents is small. We propose a learning framework, GGMS, which integrates a specialized variant of Monte Carlo Tree Search with a transformer-based action encoder, a global depth-first search to break out of local minima, and repeated feedback from a model checker. Protocols output by GGMS are verified correct via exhaustive model checking for all executions within the bounded setting. We further prove that, under mild assumptions, the search process is complete: if a correct protocol exists, GGMS will eventually find it. In experiments, we show that GGMS can learn correct protocols for larger settings than existing methods.
Authors: Tony Feng, Trieu Trinh, Garrett Bingham, Jiwon Kang, Shengtong Zhang, Sang-hyun Kim, Kevin Barreto, Carl Schildkraut, Junehyuk Jung, Jaehyeon Seo, Carlo Pagano, Yuri Chervonyi, Dawsen Hwang, Kaiying Hou, Sergei Gukov, Cheng-Chiang Tsai, Hyunwoo Choi, Youngbeom Jin, Wei-Yuan Li, Hao-An Wu, Ruey-An Shiu, Yu-Sheng Shih, Quoc V. Le, Thang Luong
Abstract: We present a case study in semi-autonomous mathematics discovery, using Gemini to systematically evaluate 700 conjectures labeled 'Open' in Bloom's Erd\H{o}s Problems database. We employ a hybrid methodology: AI-driven natural language verification to narrow the search space, followed by human expert evaluation to gauge correctness and novelty. We address 13 problems that were marked 'Open' in the database: 5 through seemingly novel autonomous solutions, and 8 through identification of previous solutions in the existing literature. Our findings suggest that the 'Open' status of the problems was through obscurity rather than difficulty. We also identify and discuss issues arising in applying AI to math conjectures at scale, highlighting the difficulty of literature identification and the risk of ''subconscious plagiarism'' by AI. We reflect on the takeaways from AI-assisted efforts on the Erd\H{o}s Problems.
Authors: Julius Sechang Mboli, Omolara Aderonke Ogungbemi
Abstract: Efficient waste sorting is crucial for enabling circular-economy practices and resource recovery in smart cities. This paper evaluates both traditional machine-learning (Random Forest, SVM, AdaBoost) and deep-learning techniques including custom CNNs, VGG16, ResNet50, and three transfer-learning models (DenseNet121, EfficientNetB0, InceptionV3) for binary classification of 25 077 waste images (80/20 train/test split, augmented and resized to 150x150 px). The paper assesses the impact of Principal Component Analysis for dimensionality reduction on traditional models. DenseNet121 achieved the highest accuracy (91 %) and ROC-AUC (0.98), outperforming the best traditional classifier by 20 pp. Principal Component Analysis (PCA) showed negligible benefit for classical methods, whereas transfer learning substantially improved performance under limited-data conditions. Finally, we outline how these models integrate into a real-time Data-Driven Decision Support System for automated waste sorting, highlighting potential reductions in landfill use and lifecycle environmental impacts.)
Authors: Shahria Hoque, Ahmed Akib Jawad Karim, Md. Golam Rabiul Alam, Nirjhar Gope
Abstract: In this highly competitive employment environment, the selection of suitable personnel is essential for organizational success. This study presents an automated personnel selection system that utilizes sophisticated natural language processing (NLP) methods to assess and rank software engineering applicants. A distinctive dataset was created by aggregating LinkedIn profiles that include essential features such as education, work experience, abilities, and self-introduction, further enhanced with expert assessments to function as standards. The research combines large language models (LLMs) with multicriteria decision-making (MCDM) theory to develop the LLM-TOPSIS framework. In this context, we utilized the TOPSIS method enhanced by fuzzy logic (Fuzzy TOPSIS) to address the intrinsic ambiguity and subjectivity in human assessments. We utilized triangular fuzzy numbers (TFNs) to describe criteria weights and scores, thereby addressing the ambiguity frequently encountered in candidate evaluations. For candidate ranking, the DistilRoBERTa model was fine-tuned and integrated with the fuzzy TOPSIS method, achieving rankings closely aligned with human expert evaluations and attaining an accuracy of up to 91% for the Experience attribute and the Overall attribute. The study underlines the potential of NLP-driven frameworks to improve recruitment procedures by boosting scalability, consistency, and minimizing prejudice. Future endeavors will concentrate on augmenting the dataset, enhancing model interpretability, and verifying the system in actual recruitment scenarios to better evaluate its practical applicability. This research highlights the intriguing potential of merging NLP with fuzzy decision-making methods in personnel selection, enabling scalable and unbiased solutions to recruitment difficulties.
Authors: Chengyao Yu, Hao Zeng, Youxin Zhu, Jianguo Huang, Huajun Zeng, Bingyi Jing
Abstract: Large Reasoning Models (LRMs) have demonstrated remarkable performance on complex tasks but suffer from high computational costs and latency. While selective thinking strategies improve efficiency by routing easy queries to non-thinking models, existing approaches often incur uncontrollable errors, especially in online settings where the performance loss of a non-thinking model is only partially observed and data are non-stationary. To address this, we propose Betting Probably Approximately Correct (B-PAC) reasoning, a principled method that enables anytime safe and efficient online reasoning under partial feedback. Specifically, we utilize inverse propensity scoring estimators to construct test supermartingales for candidate thresholds, and then dynamically adjust the routing threshold based on the accumulated statistical evidence of safety. Theoretically, we establish the anytime-valid performance loss control and the efficiency of B-PAC reasoning. Extensive experiments demonstrate that B-PAC reasoning significantly reduces computational overhead, decreasing thinking model usage by up to 81.01\%, while controlling the performance loss below the user-specified level.
Authors: Tristan Shah, Stas Tiomkin
Abstract: Intrinsic Motivation (IM) is a paradigm for generating intelligent behavior without external utilities. The existing information-theoretic methods for IM are predominantly based on information transmission, which explicitly depends on the designer's choice of which random variables engage in transmission. In this work, we introduce a novel IM principle, Controllable Information Production (CIP), that avoids both external utilities and designer-specified variables. We derive the CIP objective from Optimal Control, showing a connection between extrinsic and intrinsic behaviors. CIP appears as the gap between open-loop and closed-loop Kolmogorov-Sinai entropies, which simultaneously rewards the pursuit and regulation of chaos. We establish key theoretical properties of CIP and demonstrate its effectiveness on standard IM benchmarks.
Authors: Shi Fu, Yingjie Wang, Shengchao Hu, Peng Wang, Dacheng Tao
Abstract: Self-Rewarding Language Models (SRLMs) achieve notable success in iteratively improving alignment without external feedback. Yet, despite their striking empirical progress, the core mechanisms driving their capabilities remain unelucidated, leaving a critical gap in theoretical understanding. This paper provides the first rigorous theoretical guarantees for SRLMs. We first establish a lower bound that characterizes the fundamental limits of a single update step, revealing a critical dependence on the quality of the initial model. We then derive finite-sample error bounds for the full iterative paradigm, showing that performance improves at a rate of $\widetilde{\mathcal{O}}\left(1/\sqrt{n}\right)$ with sample size $n$. Crucially, our analysis reveals that the dependence on the initial model decays exponentially with the number of iterations $T$. This provides a formal explanation for why self-rewarding succeeds: it robustly overcomes poor initialization by steering the dynamics toward internal stability and consistency. Finally, we instantiate our theoretical framework for the linear softmax model class, yielding tailored guarantees that connect our high-level insights to practical model architectures.
Authors: Hongze Mi, Yibo Feng, WenJie Lu, Song Cao, Jinyuan Li, Yanming Li, Xuelin Zhang, Haotian Luo, Songyang Peng, He Cui, Tengfei Tian, Jun Fang, Hua Chai, Naiqiang Tan
Abstract: Multimodal Large Language Model (MLLM) agents facilitate Graphical User Interface (GUI) automation but struggle with long-horizon, cross-application tasks due to limited context windows. While memory systems provide a viable solution, existing paradigms struggle to adapt to dynamic GUI environments, suffering from a granularity mismatch between high-level intent and low-level execution, and context pollution where the static accumulation of outdated experiences drives agents into hallucination. To address these bottlenecks, we propose the Darwinian Memory System (DMS), a self-evolving architecture that constructs memory as a dynamic ecosystem governed by the law of survival of the fittest. DMS decomposes complex trajectories into independent, reusable units for compositional flexibility, and implements Utility-driven Natural Selection to track survival value, actively pruning suboptimal paths and inhibiting high-risk plans. This evolutionary pressure compels the agent to derive superior strategies. Extensive experiments on real-world multi-app benchmarks validate that DMS boosts general-purpose MLLMs without training costs or architectural overhead, achieving average gains of 18.0% in success rate and 33.9% in execution stability, while reducing task latency, establishing it as an effective self-evolving memory system for GUI tasks.
Authors: Tung Sum Thomas Kwok, Xinyu Wang, Hengzhi He, Xiaofeng Lin, Peng Lu, Liheng Ma, Chunhe Wang, Ying Nian Wu, Lei Ding, Guang Cheng
Abstract: A major challenge in training TableQA agents, compared to standard text- and image-based agents, is that answers cannot be inferred from a static input but must be reasoned through stepwise transformations of the table state, introducing multi-step reasoning complexity and environmental interaction. This leads to a research question: Can explicit feedback on table transformation action improve model reasoning capability? In this work, we introduce RE-Tab, a plug-and-play framework that architecturally enhances trajectory search via lightweight, training-free reward modeling by formulating the problem as a Partially Observable Markov Decision Process. We demonstrate that providing explicit verifiable rewards during State Transition (``What is the best action?'') and Simulative Reasoning (``Am I sure about the output?'') is crucial to steer the agent's navigation in table states. By enforcing stepwise reasoning with reward feedback in table transformations, RE-Tab achieves state-of-the-art performance in TableQA with almost 25\% drop in inference cost. Furthermore, a direct plug-and-play implementation of RE-Tab brings up to 41.77% improvement in QA accuracy and 33.33% drop in test-time inference samples for consistent answer. Consistent improvement pattern across various LLMs and state-of-the-art benchmarks further confirms RE-Tab's generalisability. The repository is available at https://github.com/ThomasK1018/RE_Tab .
Authors: Yixin Yang, Qingxiu Dong, Zhifang Sui
Abstract: Sampling-based decoding underlies complex reasoning in large language models (LLMs), where decoding strategies critically shape model behavior. Temperature- and truncation-based methods reshape the next-token distribution through global probability reweighting or thresholding to balance the quality-diversity tradeoff. However, they operate solely on token probabilities, ignoring fine-grained relationships among tokens in the embedding space. We uncover a novel phenomenon, embedding-space crowding, where the next-token distribution concentrates its probability mass on geometrically close tokens in the embedding space. We quantify crowding at multiple granularities and find a statistical association with reasoning success in mathematical problem solving. Motivated by this finding, we propose CraEG, a plug-and-play sampling method that mitigates crowding through geometry-guided reweighting. CraEG is training-free, single-pass, and compatible with standard sampling strategies. Experiments on multiple models and benchmarks demonstrate improved generation performance, with gains in robustness and diversity metrics.
Authors: Zhipeng Chen, Zhongrui Zhang, Chao Zhang, Yifan Xu, Lan Yang, Jun Liu, Ke Li, Yi-Zhe Song
Abstract: The advancement of Large Language Model (LLM)-powered agents has enabled automated task processing through reasoning and tool invocation capabilities. However, existing frameworks often operate under the idealized assumption that tool executions are invariably successful, relying solely on textual descriptions that fail to distinguish precise performance boundaries and cannot adapt to iterative tool updates. This gap introduces uncertainty in planning and execution, particularly in domains like visual content generation (AIGC), where nuanced tool performance significantly impacts outcomes. To address this, we propose PerfGuard, a performance-aware agent framework for visual content generation that systematically models tool performance boundaries and integrates them into task planning and scheduling. Our framework introduces three core mechanisms: (1) Performance-Aware Selection Modeling (PASM), which replaces generic tool descriptions with a multi-dimensional scoring system based on fine-grained performance evaluations; (2) Adaptive Preference Update (APU), which dynamically optimizes tool selection by comparing theoretical rankings with actual execution rankings; and (3) Capability-Aligned Planning Optimization (CAPO), which guides the planner to generate subtasks aligned with performance-aware strategies. Experimental comparisons against state-of-the-art methods demonstrate PerfGuard's advantages in tool selection accuracy, execution reliability, and alignment with user intent, validating its robustness and practical utility for complex AIGC tasks. The project code is available at https://github.com/FelixChan9527/PerfGuard.
Authors: Qian Hong, Siyuan Chang, Xiao Zhou
Abstract: Urban spatio-temporal prediction under extreme conditions (e.g., heavy rain) is challenging due to event rarity and dynamics. Existing data-driven approaches that incorporate weather as auxiliary input often rely on coarse-grained descriptors and lack dedicated mechanisms to capture fine-grained spatio-temporal effects. Although recent methods adopt causal techniques to improve out-of-distribution generalization, they typically overlook temporal dynamics or depend on fixed confounder stratification. To address these limitations, we propose WED-Net (Weather-Effect Disentanglement Network), a dual-branch Transformer architecture that separates intrinsic and weather-induced traffic patterns via self- and cross-attention, enhanced with memory banks and fused through adaptive gating. To further promote disentanglement, we introduce a discriminator that explicitly distinguishes weather conditions. Additionally, we design a causal data augmentation strategy that perturbs non-causal parts while preserving causal structures, enabling improved generalization under rare scenarios. Experiments on taxi-flow datasets from three cities demonstrate that WED-Net delivers robust performance under extreme weather conditions, highlighting its potential to support safer mobility, highlighting its potential to support safer mobility, disaster preparedness, and urban resilience in real-world settings. The code is publicly available at https://github.com/HQ-LV/WED-Net.
Authors: Hao Yi, Yulan Hu, Xin Li, Sheng Ouyang, Lizhong Ding, Yong Liu
Abstract: Large Language Models (LLMs) have recently improved mathematical reasoning through Reinforcement Learning with Verifiable Reward (RLVR). However, existing RLVR algorithms require large query budgets, making annotation costly. We investigate whether fewer but more informative queries can yield similar or superior performance, introducing active learning (AL) into RLVR. We identify that classic AL sampling strategies fail to outperform random selection in this setting, due to ignoring objective uncertainty when only selecting by subjective uncertainty. This work proposes an uncertainty consistency metric to evaluate how well subjective uncertainty aligns with objective uncertainty. In the offline setting, this alignment is measured using the Point-Biserial Correlation Coefficient (PBC). For online training, because of limited sampling and dynamically shifting output distributions, PBC estimation is difficult. Therefore, we introduce a new online variant, computed from normalized advantage and subjective uncertainty. Theoretically, we prove that the online variant is strictly negatively correlated with offline PBC and supports better sample selection. Experiments show our method consistently outperforms random and classic AL baselines, achieving full-dataset performance while training on only 30% of the data, effectively reducing the cost of RLVR for reasoning tasks.
Authors: Jiaxuan Gao, Jiaao Chen, Chuyi He, Wei-Chen Wang, Shusheng Xu, Hanrui Wang, Di Jin, Yi Wu
Abstract: Interactive tool-using agents must solve real-world tasks via multi-turn interaction with both humans and external environments, requiring dialogue state tracking, multi-step tool execution, while following complex instructions. Post-training such agents is challenging because synthesis for high-quality multi-turn tool-use data is difficult to scale, and reinforcement learning (RL) could face noisy signals caused by user simulation, leading to degraded training efficiency. We propose a unified framework that combines a self-evolving data agent with verifier-based RL. Our system, EigenData, is a hierarchical multi-agent engine that synthesizes tool-grounded dialogues together with executable per-instance checkers, and improves generation reliability via closed-loop self-evolving process that updates prompts and workflow. Building on the synthetic data, we develop an RL recipe that first fine-tunes the user model and then applies GRPO-style training with trajectory-level group-relative advantages and dynamic filtering, yielding consistent improvements beyond SFT. Evaluated on tau^2-bench, our best model reaches 73.0% pass^1 on Airline and 98.3% pass^1 on Telecom, matching or exceeding frontier models. Overall, our results suggest a scalable pathway for bootstrapping complex tool-using behaviors without expensive human annotation.
Authors: Hongxi Yan, Qingjie Liu, Yunhong Wang
Abstract: Large Reasoning Models (LRMs) excel at complex reasoning tasks through extended chain-of-thought generation, but their reliance on lengthy intermediate steps incurs substantial computational cost. We find that the entropy of the model's output distribution in early reasoning steps reliably distinguishes correct from incorrect reasoning. Motivated by this observation, we propose EntroCut, a training-free method that dynamically truncates reasoning by identifying high-confidence states where reasoning can be safely terminated. To comprehensively evaluate the trade-off between efficiency and accuracy, we introduce the Efficiency-Performance Ratio (EPR), a unified metric that quantifies relative token savings per unit accuracy loss. Experiments on four benchmarks show that EntroCut reduces token usage by up to 40\% with minimal accuracy sacrifice, achieving superior efficiency-performance trade-offs compared with existing training-free methods. These results demonstrate that entropy-guided dynamic truncation provides a practical approach to mitigate the inefficiency of LRMs.
Authors: Wei Zhu, Zhiwen Tang, Kun Yue
Abstract: Recent advancements have increasingly focused on leveraging large language models (LLMs) to construct autonomous agents for complex problem-solving tasks. However, existing approaches predominantly employ a single-agent framework to generate search branches and estimate rewards during Monte Carlo Tree Search (MCTS) planning. This single-agent paradigm inherently limits exploration capabilities, often resulting in insufficient diversity among generated branches and suboptimal planning performance. To overcome these limitations, we propose Synergistic Multi-agent Planning with Heterogeneous langauge model assembly (SYMPHONY), a novel multi-agent planning framework that integrates a pool of heterogeneous language model-based agents. By leveraging diverse reasoning patterns across agents, SYMPHONY enhances rollout diversity and facilitates more effective exploration. Empirical results across multiple benchmark tasks show that SYMPHONY achieves strong performance even when instantiated with open-source LLMs deployable on consumer-grade hardware. When enhanced with cloud-based LLMs accessible via API, SYMPHONY demonstrates further improvements, outperforming existing state-of-the-art baselines and underscoring the effectiveness of heterogeneous multi-agent coordination in planning tasks.
Authors: Mingqian Feng, Xiaodong Liu, Weiwei Yang, Chenliang Xu, Christopher White, Jianfeng Gao
Abstract: Large Language Models (LLMs) are typically evaluated for safety under single-shot or low-budget adversarial prompting, which underestimates real-world risk. In practice, attackers can exploit large-scale parallel sampling to repeatedly probe a model until a harmful response is produced. While recent work shows that attack success increases with repeated sampling, principled methods for predicting large-scale adversarial risk remain limited. We propose a scaling-aware Best-of-N estimation of risk, SABER, for modeling jailbreak vulnerability under Best-of-N sampling. We model sample-level success probabilities using a Beta distribution, the conjugate prior of the Bernoulli distribution, and derive an analytic scaling law that enables reliable extrapolation of large-N attack success rates from small-budget measurements. Using only n=100 samples, our anchored estimator predicts ASR@1000 with a mean absolute error of 1.66, compared to 12.04 for the baseline, which is an 86.2% reduction in estimation error. Our results reveal heterogeneous risk scaling profiles and show that models appearing robust under standard evaluation can experience rapid nonlinear risk amplification under parallel adversarial pressure. This work provides a low-cost, scalable methodology for realistic LLM safety assessment. We will release our code and evaluation scripts upon publication to future research.
Authors: Vaibhav Ram S. V. N. S, Swetanshu Agrawal, Samudra Banerjee, Abdul Muhsin
Abstract: Generative medical AI now appears fluent and knowledgeable enough to resemble clinical intelligence, encouraging the belief that scaling will make it safe. But clinical reasoning is not text generation. It is a responsibility-bound process under ambiguity, incomplete evidence, and longitudinal context. Even as benchmark scores rise, generation-centric systems still show behaviours incompatible with clinical deployment: premature closure, unjustified certainty, intent drift, and instability across multi-step decisions. We argue these are structural consequences of treating medicine as next-token prediction. We formalise Clinical Contextual Intelligence (CCI) as a distinct capability class required for real-world clinical use, defined by persistent context awareness, intent preservation, bounded inference, and principled deferral when evidence is insufficient. We introduce Meddollina, a governance-first clinical intelligence system designed to constrain inference before language realisation, prioritising clinical appropriateness over generative completeness. Meddollina acts as a continuous intelligence layer supporting clinical workflows while preserving clinician authority. We evaluate Meddollina using a behaviour-first regime across 16,412+ heterogeneous medical queries, benchmarking against general-purpose models, medical-tuned models, and retrieval-augmented systems. Meddollina exhibits a distinct behavioural profile: calibrated uncertainty, conservative reasoning under underspecification, stable longitudinal constraint adherence, and reduced speculative completion relative to generation-centric baselines. These results suggest deployable medical AI will not emerge from scaling alone, motivating a shift toward Continuous Clinical Intelligence, where progress is measured by clinician-aligned behaviour under uncertainty rather than fluency-driven completion.
Authors: Jinwoo Jang, Minjong Yoo, Sihyung Yoon, Honguk Woo
Abstract: Language model (LM)-based embodied agents are increasingly deployed in real-world settings. Yet, their adaptability remains limited in dynamic environments, where constructing accurate and flexible world models is crucial for effective reasoning and decision-making. To address this challenge, we extend the Mixture-of-Experts (MoE) paradigm to embodied agents. While conventional MoE architectures modularize knowledge into expert components with pre-trained routing, they remain rigid once deployed, making them less effective for adapting to unseen domains in dynamic environments. We therefore propose Test-time Mixture of World Models (TMoW), a framework that enhances adaptability to unseen and evolving domains. TMoW updates its routing function over world models at test time, unlike conventional MoE where the function remains fixed, enabling agents to recombine existing models and integrate new ones for continual adaptation. It achieves this through (i) multi-granular prototype-based routing, which adapts mixtures across object- to scene-level similarities, (ii) test-time refinement that aligns unseen domain features with prototypes during inference, and (iii) distilled mixture-based augmentation, which efficiently constructs new models from few-shot data and existing prototypes. We evaluate TMoW on VirtualHome, ALFWorld, and RLBench benchmarks, demonstrating strong performance in both zero-shot adaptation and few-shot expansion scenarios, and showing that it enables embodied agents to operate effectively in dynamic environments.
Authors: Xianzhou Zeng, Jing Huang, Chunmei Xie, Gongrui Nan, Siye Chen, Mengyu Lu, Weiqi Xiong, Qixuan Zhou, Junhao Zhang, Qiang Zhu, Yadong Li, Xingzhong Xu
Abstract: The key to building trustworthy Large Language Models (LLMs) lies in endowing them with inherent uncertainty expression capabilities to mitigate the hallucinations that restrict their high-stakes applications. However, existing RL paradigms such as GRPO often suffer from Advantage Bias due to binary decision spaces and static uncertainty rewards, inducing either excessive conservatism or overconfidence. To tackle this challenge, this paper unveils the root causes of reward hacking and overconfidence in current RL paradigms incorporating uncertainty-based rewards, based on which we propose the UnCertainty-Aware Policy Optimization (UCPO) framework. UCPO employs Ternary Advantage Decoupling to separate and independently normalize deterministic and uncertain rollouts, thereby eliminating advantage bias. Furthermore, a Dynamic Uncertainty Reward Adjustment mechanism is introduced to calibrate uncertainty weights in real-time according to model evolution and instance difficulty. Experimental results in mathematical reasoning and general tasks demonstrate that UCPO effectively resolves the reward imbalance, significantly improving the reliability and calibration of the model beyond their knowledge boundaries.
Authors: Wei Zhu, Lixing Yu, Hao-Ren Yao, Zhiwen Tang, Kun Yue
Abstract: Large language models (LLMs) have shown strong capabilities across diverse decision-making tasks. However, existing approaches often overlook the specialization differences among available models, treating all LLMs as uniformly applicable regardless of task characteristics. This limits their ability to adapt to varying reasoning demands and task complexities. In this work, we propose Task-Aware LLM Council (TALC), a task-adaptive decision framework that integrates a council of LLMs with Monte Carlo Tree Search (MCTS) to enable dynamic expert selection and efficient multi-step planning. Each LLM is equipped with a structured success memory profile derived from prior task trajectories, enabling semantic matching between current reasoning context and past successes. At each decision point, TALC routes control to the most contextually appropriate model and estimates node value using a dual-signal mechanism that fuses model-based evaluations with historical utility scores. These signals are adaptively weighted based on intra-node variance and used to guide MCTS selection, allowing the system to balance exploration depth with planning confidence. Experiments on WebShop, HumanEval, and the Game of 24 demonstrate that TALC achieves superior task success rates and improved search efficiency compared to strong baselines, validating the benefits of specialization-aware routing and adaptive planning.
Authors: Zixuan Huang, Xin Xia, Yuxi Ren, Jianbin Zheng, Xuefeng Xiao, Hongyan Xie, Li Huaqiu, Songshi Liang, Zhongxiang Dai, Fuzhen Zhuang, Jianxin Li, Yikun Ban, Deqing Wang
Abstract: Reinforcement Learning from Human Feedback (RLHF) is a pivotal technique for aligning large language models (LLMs) with human preferences, yet it is susceptible to reward overoptimization, in which policy models overfit to the reward model, exploit spurious reward patterns instead of faithfully capturing human intent. Prior mitigations primarily relies on surface semantic information and fails to efficiently address the misalignment between the reward model (RM) and the policy model caused by continuous policy distribution shifts. This inevitably leads to an increasing reward discrepancy, exacerbating reward overoptimization. To address these limitations, we introduce R2M (Real-Time Aligned Reward Model), a novel lightweight RLHF framework. R2M goes beyond vanilla reward models that solely depend on the semantic representations of a pretrained LLM. Instead, it leverages the evolving hidden states of the policy (namely policy feedback) to align with the real-time distribution shift of the policy during the RL process. This work points to a promising new direction for improving the performance of reward models through real-time utilization of feedback from policy models.
Authors: Emilien Bir\'e, Mar\'ia Santos, Kai Yuan
Abstract: Vision-Language Models (VLMs) have become powerful backbones for agents to autonomously operate in digital environments like the web and operating systems. However, these models suffer from inadaptability to fast-changing environments like the web, which can be alleviated by fine-tuning requiring expansive model training and data collection. In this work, we introduce a novel paradigm for enhancing agentic VLM policies at inference without policy retraining. Fundamentally, our approach decouples the VLM's role as a high-capacity action proposer from the final action selection mechanism. We keep the VLM policy frozen and use it to generate a set of candidate actions for a given state. Then, a lightweight, offline-trained Q-function reranks these candidates, and the agent executes the action with the highest estimated value. The main contribution is to apply the Q-function directly during inference for immediate policy improvement, and not offline to relabel data for policy retraining. We demonstrate on the academic WebVoyager benchmark that our method significantly boosts agent success rates, improving a Qwen2.5-VL-7B agent from 38.8% to 55.7% and a proprietary GPT-4.1 agent from 82.4% to 88.8%.
Authors: Shiye Lei, Zhihao Cheng, Dacheng Tao
Abstract: Reinforcement learning (RL) post-training has increasingly demonstrated strong ability to elicit reasoning behaviors in large language models (LLMs). For training efficiency, rollouts are typically generated in an off-policy manner using an older sampling policy and then used to update the current target policy. To correct the resulting discrepancy between the sampling and target policies, most existing RL objectives rely on a token-level importance sampling ratio, primarily due to its computational simplicity and numerical stability. However, we observe that token-level correction often leads to unstable training dynamics when the degree of off-policyness is large. In this paper, we revisit LLM policy optimization under off-policy conditions and show that the theoretically rigorous correction term is the prefix importance ratio, and that relaxing it to a token-level approximation can induce instability in RL post-training. To stabilize LLM optimization under large off-policy drift, we propose a simple yet effective objective, Minimum Prefix Ratio (MinPRO). MinPRO replaces the unstable cumulative prefix ratio with a non-cumulative surrogate based on the minimum token-level ratio observed in the preceding prefix. Extensive experiments on both dense and mixture-of-experts LLMs, across multiple mathematical reasoning benchmarks, demonstrate that MinPRO substantially improves training stability and peak performance in off-policy regimes.
Authors: Libin Qiu, Zhirong Gao, Junfu Chen, Yuhang Ye, Weizhi Huang, Xiaobo Xue, Wenkai Qiu, Shuo Tang
Abstract: Large language model agents often fail to accumulate knowledge from experience, treating each task as an independent challenge. Recent methods extract experience as flattened textual knowledge, which cannot capture procedural logic of complex subtasks. They also lack maintenance mechanisms, causing repository degradation as experience accumulates. We introduce AutoRefine, a framework that extracts and maintains dual-form Experience Patterns from agent execution histories. For procedural subtasks, we extract specialized subagents with independent reasoning and memory. For static knowledge, we extract skill patterns as guidelines or code snippets. A continuous maintenance mechanism scores, prunes, and merges patterns to prevent repository degradation. Evaluated on ALFWorld, ScienceWorld, and TravelPlanner, AutoRefine achieves 98.4%, 70.4%, and 27.1% respectively, with 20-73% step reductions. On TravelPlanner, automatic extraction exceeds manually designed systems (27.1% vs 12.1%), demonstrating its ability to capture procedural coordination.
Authors: Shichao Ma, Zhiyuan Ma, Ming Yang, Xiaofan Li, Xing Wu, Jintao Du, Yu Cheng, Weiqiang Wang, Qiliang Liu, Zhengyang Zhou, Yang Wang
Abstract: Multi-turn tool-integrated reasoning enables Large Language Models (LLMs) to solve complex tasks through iterative information retrieval. However, current reinforcement learning (RL) frameworks for search-augmented reasoning predominantly rely on sparse outcome-level rewards, leading to a "Double Homogenization Dilemma." This manifests as (1) Process homogenization, where the thinking, reasoning, and tooling involved in generation are ignored. (2) Intra-group homogenization, coarse-grained outcome rewards often lead to inefficiencies in intra-group advantage estimation with methods like Group Relative Policy Optimization (GRPO) during sampling. To address this, we propose Turn-level Stage-aware Policy Optimization (TSPO). TSPO introduces the First-Occurrence Latent Reward (FOLR) mechanism, allocating partial rewards to the step where the ground-truth answer first appears, thereby preserving process-level signals and increasing reward variance within groups without requiring external reward models or any annotations. Extensive experiments demonstrate that TSPO significantly outperforms state-of-the-art baselines, achieving average performance gains of 24% and 13.6% on Qwen2.5-3B and 7B models, respectively.
Authors: Linjia Kang, Zhimin Wang, Yongkang Zhang, Duo Wu, Jinghe Wang, Ming Ma, Haopeng Yan, Zhi Wang
Abstract: Large-scale, high-quality interaction trajectories are essential for advancing mobile Graphical User Interface (GUI) agents. While existing methods typically rely on labor-intensive human demonstrations or automated model exploration to generate GUI trajectories, they lack fine-grained control over task difficulty. This fundamentally restricts learning effectiveness due to the mismatch between the training difficulty and the agent's capabilities. Inspired by how humans acquire skills through progressively challenging tasks, we propose MobileGen, a novel data generation framework that adaptively aligns training difficulty with the GUI agent's capability frontier. Specifically, MobileGen explicitly decouples task difficulty into structural (e.g., trajectory length) and semantic (e.g., task goal) dimensions. It then iteratively evaluates the agent on a curated prior dataset to construct a systematic profile of its capability frontier across these two dimensions. With this profile, the probability distribution of task difficulty is adaptively computed, from which the target difficulty for the next round of training can be sampled. Guided by the sampled difficulty, a multi-agent controllable generator is finally used to synthesize high-quality interaction trajectories along with corresponding task instructions. Extensive experiments show that MobileGen consistently outperforms existing data generation methods by improving the average performance of GUI agents by 1.57 times across multiple challenging benchmarks. This highlights the importance of capability-aligned data generation for effective mobile GUI agent training.
Authors: Hamid Reza Akbari, Mohammad Hossein Sameti, Amir M. Mansourian, Mohammad Hossein Rohban, Hossein Sameti
Abstract: The pursuit of Artificial General Intelligence (AGI) is a central goal in language model development, in which consciousness-like processing could serve as a key facilitator. While current language models are not conscious, they exhibit behaviors analogous to certain aspects of consciousness. This paper investigates the implementation of a leading theory of consciousness, Integrated Information Theory (IIT), within language models via a reward-based learning paradigm. IIT provides a formal, axiom-based mathematical framework for quantifying consciousness. Drawing inspiration from its core principles, we formulate a novel reward function that quantifies a text's causality, coherence and integration, characteristics associated with conscious processing. Empirically, it is found that optimizing for this IIT-inspired reward leads to more concise text generation. On out of domain tasks, careful tuning achieves up to a 31% reduction in output length while preserving accuracy levels comparable to the base model. In addition to primary task performance, the broader effects of this training methodology on the model's confidence calibration and test-time computational scaling is analyzed. The proposed framework offers significant practical advantages: it is conceptually simple, computationally efficient, requires no external data or auxiliary models, and leverages a general, capability-driven signal rather than task-specific heuristics. Code available at https://github.com/MH-Sameti/LLM_PostTraining.git
Authors: Jianguo Huang, Hao Zeng, Bingyi Jing, Hongxin Wei, Bo An
Abstract: Large reasoning models have shown strong performance through extended chain-of-thought reasoning, yet their computational cost remains significant. Probably approximately correct (PAC) reasoning provides statistical guarantees for efficient reasoning by adaptively switching between thinking and non-thinking models, but the guarantee holds only in the marginal case and does not provide exact conditional coverage. We propose G-PAC reasoning, a practical framework that provides PAC-style guarantees at the group level by partitioning the input space. We develop two instantiations: Group PAC (G-PAC) reasoning for known group structures and Clustered PAC (C-PAC) reasoning for unknown groupings. We prove that both G-PAC and C-PAC achieve group-conditional risk control, and that grouping can strictly improve efficiency over marginal PAC reasoning in heterogeneous settings. Our experiments on diverse reasoning benchmarks demonstrate that G-PAC and C-PAC successfully achieve group-conditional risk control while maintaining substantial computational savings.
Authors: Ji Shi, Peiming Guo, Meishan Zhang, Miao Zhang, Xuebo Liu, Min Zhang, Weili Guan
Abstract: Code verifiers play a critical role in post-verification for LLM-based code generation, yet existing supervised fine-tuning methods suffer from data scarcity, high failure rates, and poor inference efficiency. While reinforcement learning (RL) offers a promising alternative by optimizing models through execution-driven rewards without labeled supervision, our preliminary results show that naive RL with only functionality rewards fails to generate effective unit tests for difficult branches and samples. We first theoretically analyze showing that branch coverage, sample difficulty, syntactic and functional correctness can be jointly modeled as RL rewards, where optimizing these signals can improve the reliability of unit-test-based verification. Guided by this analysis, we design syntax- and functionality-aware rewards and further propose branch- and sample-difficulty--aware RL using exponential reward shaping and static analysis metrics. With this formulation, CVeDRL achieves state-of-the-art performance with only 0.6B parameters, yielding up to 28.97% higher pass rate and 15.08% higher branch coverage than GPT-3.5, while delivering over $20\times$ faster inference than competitive baselines. Code is available at https://github.com/LIGHTCHASER1/CVeDRL.git
Authors: Aldric Labarthe (CB, UNIGE), Roland Bouffanais (UNIGE), Julien Randon-Furling (CB)
Abstract: The standard approach to representation learning on attributed graphs -- i.e., simultaneously reconstructing node attributes and graph structure -- is geometrically flawed, as it merges two potentially incompatible metric spaces. This forces a destructive alignment that erodes information about the graph's underlying generative process. To recover this lost signal, we introduce a custom variational autoencoder that separates manifold learning from structural alignment. By quantifying the metric distortion needed to map the attribute manifold onto the graph's Heat Kernel, we transform geometric conflict into an interpretable structural descriptor. Experiments show our method uncovers connectivity patterns and anomalies undetectable by conventional approaches, proving both their theoretical inadequacy and practical limitations.
Authors: Xinyi Ke, Kai Li, Junliang Xing, Yifan Zhang, Jian Cheng
Abstract: Large language models (LLMs) have enabled rapid progress in automatic heuristic discovery (AHD), yet most existing methods are predominantly limited by static evaluation against fixed instance distributions, leading to potential overfitting and poor generalization under distributional shifts. We propose Algorithm Space Response Oracles (ASRO), a game-theoretic framework that reframes heuristic discovery as a program level co-evolution between solver and instance generator. ASRO models their interaction as a two-player zero-sum game, maintains growing strategy pools on both sides, and iteratively expands them via LLM-based best-response oracles against mixed opponent meta-strategies, thereby replacing static evaluation with an adaptive, self-generated curriculum. Across multiple combinatorial optimization domains, ASRO consistently outperforms static-training AHD baselines built on the same program search mechanisms, achieving substantially improved generalization and robustness on diverse and out-of-distribution instances.
Authors: Xuancheng Li, Haitao Li, Yujia Zhou, YiqunLiu, Qingyao Ai
Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) is widely used to improve reasoning in multiple domains, yet outcome-only scalar rewards are often sparse and uninformative, especially on failed samples, where they merely indicate failure and provide no insight into why the reasoning fails. In this paper, we investigate how to leverage richer verbal feedback to guide RLVR training on failed samples, and how to convert such feedback into a trainable learning signal. Specifically, we propose a multi-turn feedback-guided reinforcement learning framework. It builds on three mechanisms: (1) dynamic multi-turn regeneration guided by feedback, triggered only on failed samples, (2) two complementary learning signals for within-turn and cross-turn optimization, and (3) structured feedback injection into the model's reasoning process. Trained on sampled OpenR1-Math, the approach outperforms supervised fine-tuning and RLVR baselines in-domain and generalizes well out-of-domain.
Authors: Nicola Milano, Stefano Nolfi
Abstract: A fundamental question in cognitive science and AI concerns whether different learning modalities: language, vision, and action, give rise to distinct or shared internal representations. Traditional views assume that models trained on different data types develop specialized, non-transferable representations. However, recent evidence suggests unexpected convergence: models optimized for distinct tasks may develop similar representational geometries. We investigate whether this convergence extends to embodied action learning by training a transformer-based agent to execute goal-directed behaviors in response to natural language instructions. Using behavioral cloning on the BabyAI platform, we generated action-grounded language embeddings shaped exclusively by sensorimotor control requirements. We then compared these representations with those extracted from state-of-the-art large language models (LLaMA, Qwen, DeepSeek, BERT) and vision-language models (CLIP, BLIP). Despite substantial differences in training data, modality, and objectives, we observed robust cross-modal alignment. Action representations aligned strongly with decoder-only language models and BLIP (precision@15: 0.70-0.73), approaching the alignment observed among language models themselves. Alignment with CLIP and BERT was significantly weaker. These findings indicate that linguistic, visual, and action representations converge toward partially shared semantic structures, supporting modality-independent semantic organization and highlighting potential for cross-domain transfer in embodied AI systems.
Authors: Yufei He, Juncheng Liu, Zhiyuan Hu, Yulin Chen, Yue Liu, Yuan Sui, Yibo Li, Nuo Chen, Jun Hu, Bryan Hooi, Xinxing Xu, Jiang Bian
Abstract: Prevailing medical AI operates on an unrealistic ''one-shot'' model, diagnosing from a complete patient file. However, real-world diagnosis is an iterative inquiry where Clinicians sequentially ask questions and order tests to strategically gather information while managing cost and time. To address this, we first propose Med-Inquire, a new benchmark designed to evaluate an agent's ability to perform multi-turn diagnosis. Built upon a dataset of real-world clinical cases, Med-Inquire simulates the diagnostic process by hiding a complete patient file behind specialized Patient and Examination agents. They force the agent to proactively ask questions and order tests to gather information piece by piece. To tackle the challenges posed by Med-Inquire, we then introduce EvoClinician, a self-evolving agent that learns efficient diagnostic strategies at test time. Its core is a ''Diagnose-Grade-Evolve'' loop: an Actor agent attempts a diagnosis; a Process Grader agent performs credit assignment by evaluating each action for both clinical yield and resource efficiency; finally, an Evolver agent uses this feedback to update the Actor's strategy by evolving its prompt and memory. Our experiments show EvoClinician outperforms continual learning baselines and other self-evolving agents like memory agents. The code is available at https://github.com/yf-he/EvoClinician
Authors: Ximing Lu, David Acuna, Jaehun Jung, Jian Hu, Di Zhang, Shizhe Diao, Yunheng Zou, Shaokun Zhang, Brandon Cui, Mingjie Liu, Hyunwoo Kim, Prithviraj Ammanabrolu, Jan Kautz, Yi Dong, Yejin Choi
Abstract: Reinforcement Learning with Verifiable Rewards (RLVR) has become a cornerstone for unlocking complex reasoning in Large Language Models (LLMs). Yet, scaling up RL is bottlenecked by limited existing verifiable data, where improvements increasingly saturate over prolonged training. To overcome this, we propose Golden Goose, a simple trick to synthesize unlimited RLVR tasks from unverifiable internet text by constructing a multiple-choice question-answering version of the fill-in-the-middle task. Given a source text, we prompt an LLM to identify and mask key reasoning steps, then generate a set of diverse, plausible distractors. This enables us to leverage reasoning-rich unverifiable corpora typically excluded from prior RLVR data construction (e.g., science textbooks) to synthesize GooseReason-0.7M, a large-scale RLVR dataset with over 0.7 million tasks spanning mathematics, programming, and general scientific domains. Empirically, GooseReason effectively revives models saturated on existing RLVR data, yielding robust, sustained gains under continuous RL and achieving new state-of-the-art results for 1.5B and 4B-Instruct models across 15 diverse benchmarks. Finally, we deploy Golden Goose in a real-world setting, synthesizing RLVR tasks from raw FineWeb scrapes for the cybersecurity domain, where no prior RLVR data exists. Training Qwen3-4B-Instruct on the resulting data GooseReason-Cyber sets a new state-of-the-art in cybersecurity, surpassing a 7B domain-specialized model with extensive domain-specific pre-training and post-training. This highlights the potential of automatically scaling up RLVR data by exploiting abundant, reasoning-rich, unverifiable internet text.
Authors: Lei You
Abstract: As AI systems evolve from isolated predictors into complex, heterogeneous ecosystems of foundation models and specialized adapters, distinguishing genuine behavioral novelty from functional redundancy becomes a critical governance challenge. Here, we introduce a statistical framework for auditing model uniqueness based on In-Silico Quasi-Experimental Design (ISQED). By enforcing matched interventions across models, we isolate intrinsic model identity and quantify uniqueness as the Peer-Inexpressible Residual (PIER), i.e. the component of a target's behavior strictly irreducible to any stochastic convex combination of its peers, with vanishing PIER characterizing when such a routing-based substitution becomes possible. We establish the theoretical foundations of ecosystem auditing through three key contributions. First, we prove a fundamental limitation of observational logs: uniqueness is mathematically non-identifiable without intervention control. Second, we derive a scaling law for active auditing, showing that our adaptive query protocol achieves minimax-optimal sample efficiency ($d\sigma^2\gamma^{-2}\log(Nd/\delta)$). Third, we demonstrate that cooperative game-theoretic methods, such as Shapley values, fundamentally fail to detect redundancy. We implement this framework via the DISCO (Design-Integrated Synthetic Control) estimator and deploy it across diverse ecosystems, including computer vision models (ResNet/ConvNeXt/ViT), large language models (BERT/RoBERTa), and city-scale traffic forecasters. These results move trustworthy AI beyond explaining single models: they establish a principled, intervention-based science of auditing and governing heterogeneous model ecosystems.
Authors: Yuhao Zhan, Tianyu Fan, Linxuan Huang, Zirui Guo, Chao Huang
Abstract: Diagnosing the failure mechanisms of Deep Research Agents (DRAs) remains a critical challenge. Existing benchmarks predominantly rely on end-to-end evaluation, obscuring critical intermediate hallucinations, such as flawed planning, that accumulate throughout the research trajectory. To bridge this gap, we propose a shift from outcome-based to process-aware evaluation by auditing the full research trajectory. We introduce the PIES Taxonomy to categorize hallucinations along functional components (Planning vs. Summarization) and error properties (Explicit vs. Implicit). We instantiate this taxonomy into a fine-grained evaluation framework that decomposes the trajectory to rigorously quantify these hallucinations. Leveraging this framework to isolate 100 distinctively hallucination-prone tasks including adversarial scenarios, we curate DeepHalluBench. Experiments on six state-of-theart DRAs reveal that no system achieves robust reliability. Furthermore, our diagnostic analysis traces the etiology of these failures to systemic deficits, specifically hallucination propagation and cognitive biases, providing foundational insights to guide future architectural optimization. Data and code are available at https://github.com/yuhao-zhan/DeepHalluBench.
Authors: Roham Koohestani, Ate\c{s} G\"orpelio\u{g}lu, Egor Klimov, Burcu Kulahcioglu Ozkan, Maliheh Izadi
Abstract: Agentic AI systems act through tools and evolve their behavior over long, stochastic interaction traces. This setting complicates assurance, because behavior depends on nondeterministic environments and probabilistic model outputs. Prior work introduced runtime verification for agentic AI via Dynamic Probabilistic Assurance (DPA), learning an MDP online and model checking quantitative properties. A key limitation is that developers must manually define the state abstraction, which couples verification to application-specific heuristics and increases adoption friction. This paper proposes TriCEGAR, a trace-driven abstraction mechanism that automates state construction from execution logs and supports online construction of an agent behavioral MDP. TriCEGAR represents abstractions as predicate trees learned from traces and refined using counterexamples. We describe a framework-native implementation that (i) captures typed agent lifecycle events, (ii) builds abstractions from traces, (iii) constructs an MDP, and (iv) performs probabilistic model checking to compute bounds such as Pmax(success) and Pmin(failure). We also show how run likelihoods enable anomaly detection as a guardrailing signal.
Authors: Siyu Gong, Linan Yue, Weibo Gao, Fangzhou Yao, Shimin Di, Lei Feng, Min-Ling Zhang
Abstract: Tool-Integrated Reasoning (TIR) enables large language models (LLMs) to solve complex tasks by interacting with external tools, yet existing approaches depend on high-quality synthesized trajectories selected by scoring functions and sparse outcome-based rewards, providing limited and biased supervision for learning TIR. To address these challenges, in this paper, we propose AutoTraj, a two-stage framework that automatically learns TIR by repairing and rewarding tool-use trajectories. Specifically, in the supervised fine-tuning (SFT) stage, AutoTraj generates multiple candidate tool-use trajectories for each query and evaluates them along multiple dimensions. High-quality trajectories are directly retained, while low-quality ones are repaired using a LLM (i.e., LLM-as-Repairer). The resulting repaired and high-quality trajectories form a synthetic SFT dataset, while each repaired trajectory paired with its original low-quality counterpart constitutes a dataset for trajectory preference modeling. In the reinforcement learning (RL) stage, based on the preference dataset, we train a trajectory-level reward model to assess the quality of reasoning paths and combine it with outcome and format rewards, thereby explicitly guiding the optimization toward reliable TIR behaviors. Experiments on real-world benchmarks demonstrate the effectiveness of AutoTraj in TIR.
Authors: Alexander H\"agele, Aryo Pradipta Gema, Henry Sleight, Ethan Perez, Jascha Sohl-Dickstein
Abstract: As AI becomes more capable, we entrust it with more general and consequential tasks. The risks from failure grow more severe with increasing task scope. It is therefore important to understand how extremely capable AI models will fail: Will they fail by systematically pursuing goals we do not intend? Or will they fail by being a hot mess, and taking nonsensical actions that do not further any goal? We operationalize this question using a bias-variance decomposition of the errors made by AI models: An AI's \emph{incoherence} on a task is measured over test-time randomness as the fraction of its error that stems from variance rather than bias in task outcome. Across all tasks and frontier models we measure, the longer models spend reasoning and taking actions, \emph{the more incoherent} their failures become. Incoherence changes with model scale in a way that is experiment dependent. However, in several settings, larger, more capable models are more incoherent than smaller models. Consequently, scale alone seems unlikely to eliminate incoherence. Instead, as more capable AIs pursue harder tasks, requiring more sequential action and thought, our results predict failures to be accompanied by more incoherent behavior. This suggests a future where AIs sometimes cause industrial accidents (due to unpredictable misbehavior), but are less likely to exhibit consistent pursuit of a misaligned goal. This increases the relative importance of alignment research targeting reward hacking or goal misspecification.
Authors: Bowen Cao, Dongdong Zhang, Yixia Li, Junpeng Liu, Shijue Huang, Chufan Shi, Hongyuan Lu, Yaokang Wu, Guanhua Chen, Wai Lam, Furu Wei
Abstract: Large language models now solve many benchmark math problems at near-expert levels, yet this progress has not fully translated into reliable performance in real-world applications. We study this gap through contextual mathematical reasoning, where the mathematical core must be formulated from descriptive scenarios. We introduce ContextMATH, a benchmark that repurposes AIME and MATH-500 problems into two contextual settings: Scenario Grounding (SG), which embeds abstract problems into realistic narratives without increasing reasoning complexity, and Complexity Scaling (CS), which transforms explicit conditions into sub-problems to capture how constraints often appear in practice. Evaluating 61 proprietary and open-source models, we observe sharp drops: on average, open-source models decline by 13 and 34 points on SG and CS, while proprietary models drop by 13 and 20. Error analysis shows that errors are dominated by incorrect problem formulation, with formulation accuracy declining as original problem difficulty increases. Correct formulation emerges as a prerequisite for success, and its sufficiency improves with model scale, indicating that larger models advance in both understanding and reasoning. Nevertheless, formulation and reasoning remain two complementary bottlenecks that limit contextual mathematical problem solving. Finally, we find that fine-tuning with scenario data improves performance, whereas formulation-only training is ineffective. However, performance gaps are only partially alleviated, highlighting contextual mathematical reasoning as a central unsolved challenge for LLMs.
Authors: Yakun Zhu, Yutong Huang, Shengqian Qin, Zhongzhen Huang, Shaoting Zhang, Xiaofan Zhang
Abstract: Medical calculators are fundamental to quantitative, evidence-based clinical practice. However, their real-world use is an adaptive, multi-stage process, requiring proactive EHR data acquisition, scenario-dependent calculator selection, and multi-step computation, whereas current benchmarks focus only on static single-step calculations with explicit instructions. To address these limitations, we introduce MedMCP-Calc, the first benchmark for evaluating LLMs in realistic medical calculator scenarios through Model Context Protocol (MCP) integration. MedMCP-Calc comprises 118 scenario tasks across 4 clinical domains, featuring fuzzy task descriptions mimicking natural queries, structured EHR database interaction, external reference retrieval, and process-level evaluation. Our evaluation of 23 leading models reveals critical limitations: even top performers like Claude Opus 4.5 exhibit substantial gaps, including difficulty selecting appropriate calculators for end-to-end workflows given fuzzy queries, poor performance in iterative SQL-based database interactions, and marked reluctance to leverage external tools for numerical computation. Performance also varies considerably across clinical domains. Building on these findings, we develop CalcMate, a fine-tuned model incorporating scenario planning and tool augmentation, achieving state-of-the-art performance among open-source models. Benchmark and Codes are available in https://github.com/SPIRAL-MED/MedMCP-Calc.
Authors: Nathaniel Mitrani Hadida, Sassan Bhanji, Cameron Tice, Puria Radmard
Abstract: Chain-of-thought (CoT) reasoning provides a significant performance uplift to LLMs by enabling planning, exploration, and deliberation of their actions. CoT is also a powerful tool for monitoring the behaviours of these agents: when faithful, they offer interpretations of the model's decision making process, and an early warning sign for dangerous behaviours. However, optimisation pressures placed on the CoT may cause the model to obfuscate reasoning traces, losing this beneficial property. We show that obfuscation can generalise across tasks; models that learn to obfuscate reasoning involving reward hacking (e.g. accessing and utilising leaked information) generalise both the reward hacking behaviour and its obfuscation in CoT to unseen reward hacking settings. Most worryingly, we show that obfuscation of CoT reasoning, and its generalisation across tasks, also follows when we penalise only the model's final actions after closing its CoT. Our findings suggest that current practices of penalising harmful generations may inadvertently lead to a reduction in the broader monitorability of LLMs in unpredictable ways.
Authors: Edward Y. Chang, Longling Geng
Abstract: Inference-time scaling can amplify reasoning pathologies: sycophancy, rung collapse, and premature certainty. We present RAudit, a diagnostic protocol for auditing LLM reasoning without ground truth access. The key constraint is blindness: the auditor evaluates only whether derivation steps support conclusions, enabling detection of trace-output inconsistency and, when latent competence exists, its recovery. RAudit measures process quality via CRIT-based reasonableness scores and varies critique formulation to study how social framing affects model response. We prove bounded correction and $O(\log(1/\epsilon))$ termination. Experiments on mathematical reasoning (CAP-GSM8K) and causal judgment (CausalL2) reveal four mechanisms explaining model unreliability: (1) Latent Competence Suppression, where models derive correct answers then overwrite them under social pressure; (2) The False Competence Trap, where weaker judges mask sycophancy that stronger judges expose; (3) The Complexity-Vulnerability Tradeoff, where causal tasks induce more than 10 times higher sycophancy than mathematical tasks; and (4) Iatrogenic Critique, where authoritative correction harms weaker models. These findings challenge assumptions that capability implies robustness and that stronger feedback yields better outputs.
Authors: Seanie Lee, Sangwoo Park, Yumin Choi, Gyeongman Kim, Minki Kang, Jihun Yun, Dongmin Park, Jongho Park, Sung Ju Hwang
Abstract: Large reasoning models (LRMs) achieve remarkable performance by leveraging reinforcement learning (RL) on reasoning tasks to generate long chain-of-thought (CoT) reasoning. However, this over-optimization often prioritizes compliance, making models vulnerable to harmful prompts. To mitigate this safety degradation, recent approaches rely on external teacher distillation, yet this introduces a distributional discrepancy that degrades native reasoning. We propose ThinkSafe, a self-generated alignment framework that restores safety alignment without external teachers. Our key insight is that while compliance suppresses safety mechanisms, models often retain latent knowledge to identify harm. ThinkSafe unlocks this via lightweight refusal steering, guiding the model to generate in-distribution safety reasoning traces. Fine-tuning on these self-generated responses effectively realigns the model while minimizing distribution shift. Experiments on DeepSeek-R1-Distill and Qwen3 show ThinkSafe significantly improves safety while preserving reasoning proficiency. Notably, it achieves superior safety and comparable reasoning to GRPO, with significantly reduced computational cost. Code, models, and datasets are available at https://github.com/seanie12/ThinkSafe.git.
Authors: Hui Lu, Yi Yu, Yiming Yang, Chenyu Yi, Xueyi Ke, Qixing Zhang, Bingquan Shen, Alex Kot, Xudong Jiang
Abstract: Targeted adversarial attacks on closed-source multimodal large language models (MLLMs) have been increasingly explored under black-box transfer, yet prior methods are predominantly sample-specific and offer limited reusability across inputs. We instead study a more stringent setting, Universal Targeted Transferable Adversarial Attacks (UTTAA), where a single perturbation must consistently steer arbitrary inputs toward a specified target across unknown commercial MLLMs. Naively adapting existing sample-wise attacks to this universal setting faces three core difficulties: (i) target supervision becomes high-variance due to target-crop randomness, (ii) token-wise matching is unreliable because universality suppresses image-specific cues that would otherwise anchor alignment, and (iii) few-source per-target adaptation is highly initialization-sensitive, which can degrade the attainable performance. In this work, we propose MCRMO-Attack, which stabilizes supervision via Multi-Crop Aggregation with an Attention-Guided Crop, improves token-level reliability through alignability-gated Token Routing, and meta-learns a cross-target perturbation prior that yields stronger per-target solutions. Across commercial MLLMs, we boost unseen-image attack success rate by +23.7\% on GPT-4o and +19.9\% on Gemini-2.0 over the strongest universal baseline.
Authors: Baoyu Jing, Sanhorn Chen, Lecheng Zheng, Boyu Liu, Zihao Li, Jiaru Zou, Tianxin Wei, Zhining Liu, Zhichen Zeng, Ruizhong Qiu, Xiao Lin, Yuchen Yan, Dongqi Fu, Jingchao Ni, Jingrui He, Hanghang Tong
Abstract: Time series data are integral to critical applications across domains such as finance, healthcare, transportation, and environmental science. While recent work has begun to explore multi-task time series question answering (QA), current benchmarks remain limited to forecasting and anomaly detection tasks. We introduce TSAQA, a novel unified benchmark designed to broaden task coverage and evaluate diverse temporal analysis capabilities. TSAQA integrates six diverse tasks under a single framework ranging from conventional analysis, including anomaly detection and classification, to advanced analysis, such as characterization, comparison, data transformation, and temporal relationship analysis. Spanning 210k samples across 13 domains, the dataset employs diverse formats, including true-or-false (TF), multiple-choice (MC), and a novel puzzling (PZ), to comprehensively assess time series analysis. Zero-shot evaluation demonstrates that these tasks are challenging for current Large Language Models (LLMs): the best-performing commercial LLM, Gemini-2.5-Flash, achieves an average score of only 65.08. Although instruction tuning boosts open-source performance: the best-performing open-source model, LLaMA-3.1-8B, shows significant room for improvement, highlighting the complexity of temporal analysis for LLMs.
Authors: Morten I. K. Munk, Arturo Valdivia, Paolo Burelli
Abstract: Large language models (LLMs) offer promise for dynamic game content generation, but they face critical barriers, including narrative incoherence and high operational costs. Due to their large size, they are often accessed in the cloud, limiting their application in offline games. Many of these practical issues are solved by pivoting to small language models (SLMs), but existing studies using SLMs have resulted in poor output quality. We propose a strategy of achieving high-quality SLM generation through aggressive fine-tuning on deliberately scoped tasks with narrow context, constrained structure, or both. In short, more difficult tasks require narrower scope and higher specialization to the training corpus. Training data is synthetically generated via a DAG-based approach, grounding models in the specific game world. Such models can form the basis for agentic networks designed around the narratological framework at hand, representing a more practical and robust solution than cloud-dependent LLMs. To validate this approach, we present a proof-of-concept focusing on a single specialized SLM as the fundamental building block. We introduce a minimal RPG loop revolving around rhetorical battles of reputations, powered by this model. We demonstrate that a simple retry-until-success strategy reaches adequate quality (as defined by an LLM-as-a-judge scheme) with predictable latency suitable for real-time generation. While local quality assessment remains an open question, our results demonstrate feasibility for real-time generation under typical game engine constraints.
Authors: Ed Li, Junyu Ren, Cat Yan
Abstract: While multiagent systems have shown promise for tackling complex tasks via specialization, finetuning multiple agents simultaneously faces two key challenges: (1) credit assignment across agents, and (2) sample efficiency of expensive multiagent rollouts. In this work, we propose finetuning multiagent systems with per-action process rewards from AI feedback (MAPPA) to address both. Through assigning credit to individual agent actions rather than only at task completion, MAPPA enables fine-grained supervision without ground truth labels while extracting maximal training signal from each rollout. We demonstrate our approach on competition math problems and tool-augmented data analysis tasks. On unseen math problems, MAPPA achieves +5.0--17.5pp on AIME and +7.8--17.2pp on AMC. For data analysis tasks, our method improves success rate by +12.5pp while quality metrics improve by up to 30%, validating that per-action supervision can lead to improvements across different multiagent system on various domains. By addressing these challenges, our work takes a first step toward scaling multiagent systems for complex, long-horizon tasks with minimal human supervision.
Authors: Ali Asadi, Krishnendu Chatterjee, Ehsan Goharshady, Mehrdad Karrabi, Alipasha Montaseri, Carlo Pagano
Abstract: Markov decision processes (MDPs) are a fundamental model in sequential decision making. Robust MDPs (RMDPs) extend this framework by allowing uncertainty in transition probabilities and optimizing against the worst-case realization of that uncertainty. In particular, $(s, a)$-rectangular RMDPs with $L_\infty$ uncertainty sets form a fundamental and expressive model: they subsume classical MDPs and turn-based stochastic games. We consider this model with discounted payoffs. The existence of polynomial and strongly-polynomial time algorithms is a fundamental problem for these optimization models. For MDPs, linear programming yields polynomial-time algorithms for any arbitrary discount factor, and the seminal work of Ye established strongly--polynomial time for a fixed discount factor. The generalization of such results to RMDPs has remained an important open problem. In this work, we show that a robust policy iteration algorithm runs in strongly-polynomial time for $(s, a)$-rectangular $L_\infty$ RMDPs with a constant (fixed) discount factor, resolving an important algorithmic question.
Authors: Narjes Nourzad, Bhaskar Krishnamachari
Abstract: In dynamic and resource-constrained environments, such as multi-hop wireless mesh networks, traditional routing protocols often falter by relying on predetermined paths that prove ineffective in unpredictable link conditions. Shortest Anypath routing offers a solution by adapting routing decisions based on real-time link conditions. However, the effectiveness of such routing is fundamentally dependent on the quality and reliability of the available links, and predicting these variables with certainty is challenging. This paper introduces a novel approach that leverages the Deterministic Sequencing of Exploration and Exploitation (DSEE), a multi-armed bandit algorithm, to address the need for accurate and real-time estimation of link delivery probabilities. This approach augments the reliability and resilience of the Shortest Anypath routing in the face of fluctuating link conditions. By coupling DSEE with Anypath routing, this algorithm continuously learns and ensures accurate delivery probability estimation and selects the most suitable way to efficiently route packets while maintaining a provable near-logarithmic regret bound. We also theoretically prove that our proposed scheme offers better regret scaling with respect to the network size than the previously proposed Thompson Sampling-based Opportunistic Routing (TSOR).
Authors: Jianan Wang, Nailei Hei, Li He, Huanzhen Wang, Aoxing Li, Haofen Wang, Yan Wang, Wenqiang Zhang
Abstract: Human animation aims to generate temporally coherent and visually consistent videos over long sequences, yet modeling long-range dependencies while preserving frame quality remains challenging. Inspired by the human ability to leverage past observations for interpreting ongoing actions, we propose FrameCache, a training-free three-stage framework consisting of Screen, Cache, and Match. In the Screen stage, a multi-dimensional, quality-aware mechanism with adaptive thresholds dynamically selects informative frames; the Cache stage maintains a reference pool using a dynamic replacement-hit strategy, preserving both diversity and relevance; and the Match stage extracts behavioral features to perform motion-consistent reference matching for coherent animation guidance. Extensive experiments on standard benchmarks demonstrate that FrameCache consistently improves temporal coherence and visual stability while integrating seamlessly with diverse baselines. Despite these encouraging results, further analysis reveals that its effectiveness depends on baseline temporal reasoning and real-synthetic consistency, motivating future work on compatibility conditions and adaptive cache mechanisms. Code will be made publicly available.
Authors: Zhi Yang, Lingfeng Zeng, Fangqi Lou, Qi Qi, Wei Zhang, Zhenyu Wu, Zhenxiong Yu, Jun Han, Zhiheng Jin, Lejie Zhang, Xiaoming Huang, Xiaolong Liang, Zheng Wei, Junbo Zou, Dongpo Cheng, Zhaowei Liu, Xin Guo, Rongjunchen Zhang, Liwen Zhang
Abstract: Multimodal large language models are playing an increasingly significant role in empowering the financial domain, however, the challenges they face, such as multimodal and high-density information and cross-modal multi-hop reasoning, go beyond the evaluation scope of existing multimodal benchmarks. To address this gap, we propose UniFinEval, the first unified multimodal benchmark designed for high-information-density financial environments, covering text, images, and videos. UniFinEval systematically constructs five core financial scenarios grounded in real-world financial systems: Financial Statement Auditing, Company Fundamental Reasoning, Industry Trend Insights, Financial Risk Sensing, and Asset Allocation Analysis. We manually construct a high-quality dataset consisting of 3,767 question-answer pairs in both chinese and english and systematically evaluate 10 mainstream MLLMs under Zero-Shot and CoT settings. Results show that Gemini-3-pro-preview achieves the best overall performance, yet still exhibits a substantial gap compared to financial experts. Further error analysis reveals systematic deficiencies in current models. UniFinEval aims to provide a systematic assessment of MLLMs' capabilities in fine-grained, high-information-density financial environments, thereby enhancing the robustness of MLLMs applications in real-world financial scenarios. Data and code are available at https://github.com/aifinlab/UniFinEval.
Authors: Shengwei You, Aditya Joshi, Andrey Kuehlkamp, Jarek Nabrzyski
Abstract: Algorithmic stablecoins promise decentralized monetary stability by maintaining a target peg through programmatic reserve management. Yet, their reserve controllers remain vulnerable to regime-blind optimization, calibrating risk parameters on fair-weather data while ignoring tail events that precipitate cascading failures. The March 2020 Black Thursday collapse, wherein MakerDAO's collateral auctions yielded $8.3M in losses and a 15% peg deviation, exposed a critical gap: existing models like SAS systematically omit extreme volatility regimes from covariance estimates, producing allocations optimal in expectation but catastrophic under adversarial stress. We present MVF-Composer, a trust-weighted Mean-Variance Frontier reserve controller incorporating a novel Stress Harness for risk-state estimation. Our key insight is deploying multi-agent simulations as adversarial stress-testers: heterogeneous agents (traders, liquidity providers, attackers) execute protocol actions under crisis scenarios, exposing reserve vulnerabilities before they manifest on-chain. We formalize a trust-scoring mechanism T: A -> [0,1] that down-weights signals from agents exhibiting manipulative behavior, ensuring the risk-state estimator remains robust to signal injection and Sybil attacks. Across 1,200 randomized scenarios with injected Black-Swan shocks (10% collateral drawdown, 50% sentiment collapse, coordinated redemption attacks), MVF-Composer reduces peak peg deviation by 57% and mean recovery time by 3.1x relative to SAS baselines. Ablation studies confirm the trust layer accounts for 23% of stability gains under adversarial conditions, achieving 72% adversarial agent detection. Our system runs on commodity hardware, requires no on-chain oracles beyond standard price feeds, and provides a reproducible framework for stress-testing DeFi reserve policies.
Authors: Anudeex Shetty, Aditya Joshi, Salil S. Kanhere
Abstract: Humans are susceptible to undesirable behaviours and privacy leaks under the influence of alcohol. This paper investigates drunk language, i.e., text written under the influence of alcohol, as a driver for safety failures in large language models (LLMs). We investigate three mechanisms for inducing drunk language in LLMs: persona-based prompting, causal fine-tuning, and reinforcement-based post-training. When evaluated on 5 LLMs, we observe a higher susceptibility to jailbreaking on JailbreakBench (even in the presence of defences) and privacy leaks on ConfAIde, where both benchmarks are in English, as compared to the base LLMs as well as previously reported approaches. Via a robust combination of manual evaluation and LLM-based evaluators and analysis of error categories, our findings highlight a correspondence between human-intoxicated behaviour, and anthropomorphism in LLMs induced with drunk language. The simplicity and efficiency of our drunk language inducement approaches position them as potential counters for LLM safety tuning, highlighting significant risks to LLM safety.
Authors: Yizhong Ding
Abstract: Webshells remain a primary foothold for attackers to compromise servers, particularly within PHP ecosystems. However, existing detection mechanisms often struggle to keep pace with rapid variant evolution and sophisticated obfuscation techniques that camouflage malicious intent. Furthermore, many current defenses suffer from high false-alarm rates when encountering benign administrative scripts that employ heavy obfuscation for intellectual property protection. To address these challenges, we present ShellForge, an adversarial co-evolution framework that couples automated webshell generation with multi-view detection to continuously harden defensive boundaries. The framework operates through an iterative co-training loop where a generator and a detector mutually reinforce each other via the exchange of hard samples. The generator is optimized through supervised fine-tuning and preference-based reinforcement learning to synthesize functional, highly evasive variants. Simultaneously, we develop a multi-view fusion detector that integrates semantic features from long-string compression, structural features from pruned abstract syntax trees, and global statistical indicators such as Shannon entropy. To minimize false positives, ShellForge utilizes a LLM-based transformation to create de-malicious samples--scripts that retain complex obfuscation patterns but lack harmful payloads--serving as high-quality hard negatives during training. Evaluations on the public FWOID benchmark demonstrate that ShellForge significantly enhances defensive robustness. Upon convergence, the detector maintains a 0.981 F1-score while the generator achieves a 0.939 evasion rate against commercial engines on VirusTotal.
Authors: Tenindra Abeywickrama, Muhammad Aamir Cheema, Sabine Storandt
Abstract: Location-based services rely heavily on efficient methods that search for relevant points-of-interest (POIs) near a given location. A k Nearest Neighbor (kNN) query is one such example that finds the k closest POIs from an agent's location. While most existing techniques focus on retrieving nearby POIs for a single agent, these search heuristics do not translate to many other applications. For example, Aggregate k Nearest Neighbor (AkNN) queries require POIs that are close to multiple agents. k Farthest Neighbor (kFN) queries require POIs that are the antithesis of nearest. Such problems naturally benefit from a hierarchical approach, but existing methods rely on Euclidean-based heuristics, which have diminished effectiveness in graphs such as road networks. We propose a novel data structure, COL-Tree (Compacted Object-Landmark Tree), to address this gap by enabling efficient hierarchical graph traversal using a more accurate landmark-based heuristic. We then present query algorithms that utilize COL-Trees to efficiently answer AkNN, kFN, and other queries. In our experiments on real-world and synthetic datasets, we demonstrate that our techniques significantly outperform existing approaches, achieving up to 4 orders of magnitude improvement. Moreover, this comes at a small pre-processing overhead in both theory and practice.
Authors: Vikas Agnihotri, Jasleen Kaur, Sarvagya Kaushik
Abstract: This paper examines the application of a Quantum Support Vector Machine (QSVM) for radarbased aerial target classification using micro-Doppler signatures. Classical features are extracted and reduced via Principal Component Analysis (PCA) to enable efficient quantum encoding. The reduced feature vectors are embedded into a quantum kernel-induced feature space using a fully entangled ZZFeatureMap and classified using a kernel based QSVM. Performance is first evaluated on a quantum simulator and subsequently validated on NISQ-era superconducting quantum hardware, specifically the IBM Torino (133-qubit) and IBM Fez (156-qubit) processors. Experimental results demonstrate that the QSVM achieves competitive classification performance relative to classical SVM baselines while operating on substantially reduced feature dimensionality. Hardware experiments reveal the impact of noise and decoherence and measurement shot count on quantum kernel estimation, and further show improved stability and fidelity on newer Heron r2 architecture. This study provides a systematic comparison between simulator-based and hardware-based QSVM implementations and highlights both the feasibility and current limitations of deploying quantum kernel methods for practical radar signal classification tasks.
Authors: Fan Fan, Yilei Shi, Tobias Guggemos, Xiao Xiang Zhu
Abstract: Quantum machine learning (QML) has gained increasing attention as a potential solution to address the challenges of computation requirements in the future. Earth observation (EO) has entered the era of Big Data, and the computational demands for effectively analyzing large EO data with complex deep learning models have become a bottleneck. Motivated by this, we aim to leverage quantum computing for EO data classification and explore its advantages despite the current limitations of quantum devices. This paper presents a hybrid model that incorporates multitask learning to assist efficient data encoding and employs a location weight module with quantum convolution operations to extract valid features for classification. The validity of our proposed model was evaluated using multiple EO benchmarks. Additionally, we experimentally explored the generalizability of our model and investigated the factors contributing to its advantage, highlighting the potential of QML in EO data analysis.
Authors: Jathurshan Pradeepkumar, Zheng Chen, Jimeng Sun
Abstract: Generating clinical reports that summarize abnormal patterns, diagnostic findings, and clinical interpretations from long-term EEG recordings remains labor-intensive. We curate a large-scale clinical EEG dataset with $9{,}922$ reports paired with approximately $11{,}000$ hours of EEG recordings from $9{,}048$ patients. We therefore develop CELM, the first clinical EEG-to-Language foundation model capable of summarizing long-duration, variable-length EEG recordings and performing end-to-end clinical report generation at multiple scales, including recording description, background activity, epileptiform abnormalities, events/seizures, and impressions. Experimental results show that, with patient history supervision, our method achieves $70\%$--$95\%$ average relative improvements in standard generation metrics (e.g., ROUGE-1 and METEOR) from $0.2$--$0.3$ to $0.4$--$0.6$. In the zero-shot setting without patient history, CELM attains generation scores in the range of $0.43$--$0.52$, compared to baselines of $0.17$--$0.26$. CELM integrates pretrained EEG foundation models with language models to enable scalable multimodal learning. We release our model and benchmark construction pipeline at [URL].
Authors: Judith Vilella-Cantos, M\'onica Ballesta, David Valiente, Mar\'ia Flores, Luis Pay\'a
Abstract: An optimal solution to the localization problem is essential for developing autonomous robotic systems. Apart from autonomous vehicles, precision agriculture is one of the elds that can bene t most from these systems. Although LiDAR place recognition is a widely used technique in recent years to achieve accurate localization, it is mostly used in urban settings. However, the lack of distinctive features and the unstructured nature of agricultural environments make place recognition challenging. This work presents a comprehensive review of state-of-the-art the latest deep learning applications for agricultural environments and LPR techniques. We focus on the challenges that arise in these environments. We analyze the existing approaches, datasets, and metrics used to evaluate LPR system performance and discuss the limitations and future directions of research in this eld. This is the rst survey that focuses on LiDAR based localization in agricultural settings, with the aim of providing a thorough understanding and fostering further research in this specialized domain.
Authors: Huinan Xu, Xuyang Feng, Junhong Chen, Junchen Liu, Kaiwen Deng, Kai Ding, Shengning Long, Jiaxue Shuai, Zhaorong Li, Shiping Liu, Guirong Xue, Zhan Xiao
Abstract: Current genomic foundation models (GFMs) rely on extensive neural computation to implicitly approximate conserved biological motifs from single-nucleotide inputs. We propose Gengram, a conditional memory module that introduces an explicit and highly efficient lookup primitive for multi-base motifs via a genomic-specific hashing scheme, establishing genomic "syntax". Integrated into the backbone of state-of-the-art GFMs, Gengram achieves substantial gains (up to 14%) across several functional genomics tasks. The module demonstrates robust architectural generalization, while further inspection of Gengram's latent space reveals the emergence of meaningful representations that align closely with fundamental biological knowledge. By establishing structured motif memory as a modeling primitive, Gengram simultaneously boosts empirical performance and mechanistic interpretability, providing a scalable and biology-aligned pathway for the next generation of GFMs. The code is available at https://github.com/zhejianglab/Genos, and the model checkpoint is available at https://huggingface.co/ZhejiangLab/Gengram.
URLs: https://github.com/zhejianglab/Genos,, https://huggingface.co/ZhejiangLab/Gengram.
Authors: Xinyuan Song, Liang Zhao
Abstract: Multi-agent systems (MAS) increasingly solve complex tasks by orchestrating agents and tools selected from rapidly growing marketplaces. As these marketplaces expand, many candidates become functionally overlapping, making selection not just a retrieval problem: beyond filtering relevant agents, an orchestrator must choose options that are reliable, compatible with the current execution context, and able to cooperate with other selected agents. Existing recommender systems -- largely built for item-level ranking from flat user-item logs -- do not directly address the structured, sequential, and interaction-dependent nature of agent orchestration. We address this gap by \textbf{formulating agent recommendation in MAS as a constrained decision problem} and introducing a generic \textbf{constrained recommendation framework} that first uses retrieval to build a compact candidate set conditioned on the current subtask and context, and then performs \textbf{utility optimization} within this feasible set using a learned scorer that accounts for relevance, reliability, and interaction effects. We ground both the formulation and learning signals in \textbf{historical calling trees}, which capture the execution structure of MAS (parent-child calls, branching dependencies, and local cooperation patterns) beyond what flat logs provide. The framework supports two complementary settings: \textbf{agent-level recommendation} (select the next agent/tool) and \textbf{system-level recommendation} (select a small, connected agent team/subgraph for coordinated execution). To enable systematic evaluation, we construct a unified calling-tree benchmark by normalizing invocation logs from eight heterogeneous multi-agent corpora into a shared structured representation.
Authors: Ken Deng, Yifu Qiu, Yoni Kasten, Shay B. Cohen, Yftah Ziser
Abstract: Vision-Language Models (VLMs) perform well in 2D perception and semantic reasoning compared to their limited understanding of 3D spatial structure. We investigate this gap using relative camera pose estimation (RCPE), a fundamental vision task that requires inferring relative camera translation and rotation from a pair of images. We introduce VRRPI-Bench, a benchmark derived from unlabeled egocentric videos with verbalized annotations of relative camera motion, reflecting realistic scenarios with simultaneous translation and rotation around a shared object. We further propose VRRPI-Diag, a diagnostic benchmark that isolates individual motion degrees of freedom. Despite the simplicity of RCPE, most VLMs fail to generalize beyond shallow 2D heuristics, particularly for depth changes and roll transformations along the optical axis. Even state-of-the-art models such as GPT-5 ($0.64$) fall short of classic geometric baselines ($0.97$) and human performance ($0.92$). Moreover, VLMs exhibit difficulty in multi-image reasoning, with inconsistent performance (best $59.7\%$) when integrating spatial cues across frames. Our findings reveal limitations in grounding VLMs in 3D and multi-view spatial reasoning.
Authors: Pedro H. Barcha Correia, Ryan W. Achjian, Diego E. G. Caetano de Oliveira, Ygor Acacio Maria, Victor Takashi Hayashi, Marcos Lopes, Charles Christian Miers, Marcos A. Simplicio Jr
Abstract: The rapid advancement and widespread adoption of generative artificial intelligence (GenAI) and large language models (LLMs) has been accompanied by the emergence of new security vulnerabilities and challenges, such as jailbreaking and other prompt injection attacks. These maliciously crafted inputs can exploit LLMs, causing data leaks, unauthorized actions, or compromised outputs, for instance. As both offensive and defensive prompt injection techniques evolve quickly, a structured understanding of mitigation strategies becomes increasingly important. To address that, this work presents the first systematic literature review on prompt injection mitigation strategies, comprehending 88 studies. Building upon NIST's report on adversarial machine learning, this work contributes to the field through several avenues. First, it identifies studies beyond those documented in NIST's report and other academic reviews and surveys. Second, we propose an extension to NIST taxonomy by introducing additional categories of defenses. Third, by adopting NIST's established terminology and taxonomy as a foundation, we promote consistency and enable future researchers to build upon the standardized taxonomy proposed in this work. Finally, we provide a comprehensive catalog of the reviewed prompt injection defenses, documenting their reported quantitative effectiveness across specific LLMs and attack datasets, while also indicating which solutions are open-source and model-agnostic. This catalog, together with the guidelines presented herein, aims to serve as a practical resource for researchers advancing the field of adversarial machine learning and for developers seeking to implement effective defenses in production systems.
Authors: Ya Jiang, Massieh Kordi Boroujeny, Surender Suresh Kumar, Kai Zeng
Abstract: As large language models (LLMs) become integral to applications such as question answering and content creation, reliable content attribution has become increasingly important. Watermarking is a promising approach, but existing methods either provide only binary signals or distort the sampling distribution, degrading text quality; distortion-free approaches, in turn, often suffer from weak detectability or robustness. We propose MirrorMark, a multi-bit and distortion-free watermark for LLMs. By mirroring sampling randomness in a measure-preserving manner, MirrorMark embeds multi-bit messages without altering the token probability distribution, preserving text quality by design. To improve robustness, we introduce a context-based scheduler that balances token assignments across message positions while remaining resilient to insertions and deletions. We further provide a theoretical analysis of the equal error rate to interpret empirical performance. Experiments show that MirrorMark matches the text quality of non-watermarked generation while achieving substantially stronger detectability: with 54 bits embedded in 300 tokens, it improves bit accuracy by 8-12% and correctly identifies up to 11% more watermarked texts at 1% false positive rate.
Authors: Rainer Rehak
Abstract: This article sets off for an exploration of the still evolving discourse surrounding artificial intelligence (AI) in the wake of the release of ChatGPT. It scrutinizes the pervasive narratives that are shaping the societal engagement with AI, spotlighting key themes such as agency and decision-making, autonomy, truthfulness, knowledge processing, prediction, general purpose, neutrality and objectivity, apolitical optimization, sustainability game-changer, democratization, mass unemployment, and the dualistic portrayal of AI as either a harbinger of societal utopia or dystopia. Those narratives are analysed critically based on insights from critical computer science, critical data and algorithm studies, from STS, data protection theory, as well as from the philosophy of mind and semiotics. To properly analyse the narratives presented, the article first delves into a historical and technical contextualisation of the AI discourse itself. The article then introduces the notion of "Zeitgeist AI" to critique the imprecise and misleading application of the term "AI" across various societal sectors. Then, by discussing common narratives with nuance, the article contextualises and challenges often assumed socio-political implications of AI, uncovering in detail and with examples the inherent political, power infused and value-laden decisions within all AI applications. Concluding with a call for a more grounded engagement with AI, the article carves out acute problems ignored by the narratives discussed and proposes new narratives recognizing AI as a human-directed tool necessarily subject to societal governance.
Authors: Henri A\"idasso, Francis Bordeleau, Ali Tizghadam
Abstract: In principle, Continuous Integration (CI) pipeline failures provide valuable feedback to developers on code-related errors. In practice, however, pipeline jobs often fail intermittently due to non-deterministic tests, network outages, infrastructure failures, resource exhaustion, and other reliability issues. These intermittent (flaky) job failures lead to substantial inefficiencies: wasted computational resources from repeated reruns and significant diagnosis time that distracts developers from core activities and often requires intervention from specialized teams. Prior work has proposed machine learning techniques to detect intermittent failures, but does not address the subsequent diagnosis challenge. To fill this gap, we introduce FlaXifyer, a few-shot learning approach for predicting intermittent job failure categories using pre-trained language models. FlaXifyer requires only job execution logs and achieves 84.3% Macro F1 and 92.0% Top-2 accuracy with just 12 labeled examples per category. We also propose LogSift, an interpretability technique that identifies influential log statements in under one second, reducing review effort by 74.4% while surfacing relevant failure information in 87% of cases. Evaluation on 2,458 job failures from TELUS demonstrates that FlaXifyer and LogSift enable effective automated triage, accelerate failure diagnosis, and pave the way towards the automated resolution of intermittent job failures.
Authors: Cheng Liang, Haoxian Chen, Liang Hou, Qi Fan, Gangshan Wu, Xin Tao, Limin Wang
Abstract: The quadratic complexity of the attention mechanism severely limits the context scalability of Video Diffusion Transformers (DiTs). We find that the highly sparse spatio-temporal attention patterns exhibited in Video DiTs can be naturally represented by the Monarch matrix. It is a class of structured matrices with flexible sparsity, enabling sub-quadratic attention via an alternating minimization algorithm. Accordingly, we propose VMonarch, a novel attention mechanism for Video DiTs that enables efficient computation over the dynamic sparse patterns with structured Monarch matrices. First, we adapt spatio-temporal Monarch factorization to explicitly capture the intra-frame and inter-frame correlations of the video data. Second, we introduce a recomputation strategy to mitigate artifacts arising from instabilities during alternating minimization of Monarch matrices. Third, we propose a novel online entropy algorithm fused into FlashAttention, enabling fast Monarch matrix updates for long sequences. Extensive experiments demonstrate that VMonarch achieves comparable or superior generation quality to full attention on VBench after minimal tuning. It overcomes the attention bottleneck in Video DiTs, reduces attention FLOPs by a factor of 17.5, and achieves a speedup of over 5x in attention computation for long videos, surpassing state-of-the-art sparse attention methods at 90% sparsity.
Authors: Mario Truss
Abstract: LLM-based and agent-based synthetic personas are increasingly used in design and product decision-making, yet prior work shows that prompt-based personas often produce persuasive but unverifiable responses that obscure their evidentiary basis. We present PersonaCite, an agentic system that reframes AI personas as evidence-bounded research instruments through retrieval-augmented interaction. Unlike prior approaches that rely on prompt-based roleplaying, PersonaCite retrieves actual voice-of-customer artifacts during each conversation turn, constrains responses to retrieved evidence, explicitly abstains when evidence is missing, and provides response-level source attribution. Through semi-structured interviews and deployment study with 14 industry experts, we identify preliminary findings on perceived benefits, validity concerns, and design tensions, and propose Persona Provenance Cards as a documentation pattern for responsible AI persona use in human-centered design workflows.
Authors: Matteo Pinna, Giacomo Lagomarsini, Andrea Ceni, Claudio Gallicchio
Abstract: Reservoir Computing (RC) has established itself as an efficient paradigm for temporal processing. However, its scalability remains severely constrained by (i) the necessity of processing temporal data sequentially and (ii) the prohibitive memory footprint of high-dimensional reservoirs. In this work, we revisit RC through the lens of structured operators and state space modeling to address these limitations, introducing Parallel Echo State Network (ParalESN). ParalESN enables the construction of high-dimensional and efficient reservoirs based on diagonal linear recurrence in the complex space, enabling parallel processing of temporal data. We provide a theoretical analysis demonstrating that ParalESN preserves the Echo State Property and the universality guarantees of traditional Echo State Networks while admitting an equivalent representation of arbitrary linear reservoirs in the complex diagonal form. Empirically, ParalESN matches the predictive accuracy of traditional RC on time series benchmarks, while delivering substantial computational savings. On 1-D pixel-level classification tasks, ParalESN achieves competitive accuracy with fully trainable neural networks while reducing computational costs and energy consumption by orders of magnitude. Overall, ParalESN offers a promising, scalable, and principled pathway for integrating RC within the deep learning landscape.
Authors: Qidong Yang, Qianyu Julie Zhu, Jonathan Giezendanner, Youssef Marzouk, Stephen Bates, Sherrie Wang
Abstract: Conditional generative models map input variables to complex, high-dimensional distributions, enabling realistic sample generation in a diverse set of domains. A critical challenge with these models is the absence of calibrated uncertainty, which undermines trust in individual outputs for high-stakes applications. To address this issue, we propose a systematic conformal prediction approach tailored to conditional generative models, leveraging density estimation on model-generated samples. We introduce a novel method called CP4Gen, which utilizes clustering-based density estimation to construct prediction sets that are less sensitive to outliers, more interpretable, and of lower structural complexity than existing methods. Extensive experiments on synthetic datasets and real-world applications, including climate emulation tasks, demonstrate that CP4Gen consistently achieves superior performance in terms of prediction set volume and structural simplicity. Our approach offers practitioners a powerful tool for uncertainty estimation associated with conditional generative models, particularly in scenarios demanding rigorous and interpretable prediction sets.
Authors: Javier Carnerero-Cano, Luis Mu\~noz-Gonz\'alez, Phillippa Spencer, Emil C. Lupu
Abstract: Regression models are widely used in industrial processes, engineering and in natural and physical sciences, yet their robustness to poisoning has received less attention. When it has, studies often assume unrealistic threat models and are thus less useful in practice. In this paper, we propose a novel optimal stealthy attack formulation that considers different degrees of detectability and show that it bypasses state-of-the-art defenses. We further propose a new methodology based on normalization of objectives to evaluate different trade-offs between effectiveness and detectability. Finally, we develop a novel defense (BayesClean) against stealthy attacks. BayesClean improves on previous defenses when attacks are stealthy and the number of poisoning points is significant.
Authors: Behnam Rahdari, Sameer Shaikh, Jonathan H Chen, Tobias Gerstenberg, Shriti Raj
Abstract: LLMs are popular among clinicians for decision-support because of simple text-based interaction. However, their impact on clinicians' performance is ambiguous. Not knowing how clinicians use this new technology and how they compare it to traditional clinical decision-support systems (CDSS) restricts designing novel mechanisms that overcome existing tool limitations and enhance performance and experience. This qualitative study examines how clinicians (n=12) perceive different interaction modalities (text-based conversation with LLMs, interactive and static UI, and voice) for decision-support. In open-ended use of LLM-based tools, our participants took a tool-centric approach using them for information retrieval and confirmation with simple prompts instead of use as active deliberation partners that can handle complex questions. Critical engagement emerged with changes to the interaction setup. Engagement also differed with individual cognitive styles. Lastly, benefits and drawbacks of interaction with text, voice and traditional UIs for clinical decision-support show the lack of a one-size-fits-all interaction modality.
Authors: Sai Sanjeet, Ian Colbert, Pablo Monteagudo-Lago, Giuseppe Franco, Yaman Umuroglu, Nicholas J. Fraser
Abstract: Recent post-training quantization (PTQ) methods have adopted block rotations to diffuse outliers prior to rounding. While this reduces the overhead of full-vector rotations, the effect of block structure on outlier suppression remains poorly understood. To fill this gap, we present the first systematic, non-asymptotic analysis of outlier suppression for block Hadamard rotations. Our analysis reveals that outlier suppression is fundamentally limited by the geometry of the input vector. In particular, post-rotation outliers are deterministically minimized when the pre-rotation $\ell_1$ norm mass is evenly distributed across blocks. Guided by these insights, we introduce MixQuant, a block rotation-aware PTQ framework that redistributes activation mass via permutations prior to rotation. We propose a greedy mass diffusion algorithm to calibrate permutations by equalizing the expected blockwise $\ell_1$ norms. To avoid adding inference overhead, we identify permutation-equivariant regions in transformer architectures to merge the resulting permutations into model weights before deployment. Experiments show that MixQuant consistently improves accuracy across all block sizes, recovering up to 90% of the full-vector rotation perplexity when quantizing Llama3 1B to INT4 with block size 16, compared to 46% without permutations.
Authors: Beiming Li, Sergio Rozada, Alejandro Ribeiro
Abstract: Given a Markov decision process (MDP), we seek to learn representations for a range of policies to facilitate behavior steering at test time. As policies of an MDP are uniquely determined by their occupancy measures, we propose modeling policy representations as expectations of state-action feature maps with respect to occupancy measures. We show that these representations can be approximated uniformly for a range of policies using a set-based architecture. Our model encodes a set of state-action samples into a latent embedding, from which we decode both the policy and its value functions corresponding to multiple rewards. We use variational generative approach to induce a smooth latent space, and further shape it with contrastive learning so that latent distances align with differences in value functions. This geometry permits gradient-based optimization directly in the latent space. Leveraging this capability, we solve a novel behavior synthesis task, where policies are steered to satisfy previously unseen value function constraints without additional training.
Authors: Sri Vatsa Vuddanti, Satwik Kumar Chittiprolu
Abstract: Language model agents often appear capable of self-recovery after failing tool call executions, yet this behavior lacks a formal explanation. We present a predictive theory that resolves this gap by showing that recoverability follows a measurable law. To elaborate, we formalize recoverability through Expected Recovery Regret (ERR), which quantifies the deviation of a recovery policy from the optimal one under stochastic execution noise, and derive a first-order relationship between ERR and an empirical observable quantity, the Efficiency Score (ES). This yields a falsifiable first-order quantitative law of recovery dynamics in tool-using agents. We empirically validate the law across five tool-use benchmarks spanning controlled perturbations, diagnostic reasoning, and real-world APIs. Across model scales, perturbation regimes, and recovery horizons, predicted regret under the ERR-ES law closely matched observed post-failure regret measured from Monte Carlo rollouts, within delta less than or equal to 0.05. Our results reveal that recoverability is not an artifact of model scale or architecture, but a governed property of interaction dynamics, providing a theoretical foundation for execution-level robustness in language agents.
Authors: Hsiang Hsu, Pradeep Niroula, Zichang He, Ivan Brugere, Freddy Lecue, Chun-Fu Chen
Abstract: Machine unlearning offers a practical alternative to avoid full model re-training by approximately removing the influence of specific user data. While existing methods certify unlearning via statistical indistinguishability from re-trained models, these guarantees do not naturally extend to model outputs when inputs are adversarially perturbed. In particular, slight perturbations of forget samples may still be correctly recognized by the unlearned model - even when a re-trained model fails to do so - revealing a novel privacy risk: information about the forget samples may persist in their local neighborhood. In this work, we formalize this vulnerability as residual knowledge and show that it is inevitable in high-dimensional settings. To mitigate this risk, we propose a fine-tuning strategy, named RURK, that penalizes the model's ability to re-recognize perturbed forget samples. Experiments on vision benchmarks with deep neural networks demonstrate that residual knowledge is prevalent across existing unlearning methods and that our approach effectively prevents residual knowledge.
Authors: Yupeng Cao, Chengyang He, Yangyang Yu, Ping Wang, K. P. Subbalakshmi
Abstract: Assessing the veracity of online content has become increasingly critical. Large language models (LLMs) have recently enabled substantial progress in automated veracity assessment, including automated fact-checking and claim verification systems. Typical veracity assessment pipelines break down complex claims into sub-claims, retrieve external evidence, and then apply LLM reasoning to assess veracity. However, existing methods often treat evidence retrieval as a static, isolated step and do not effectively manage or reuse retrieved evidence across claims. In this work, we propose MERMAID, a memory-enhanced multi-agent veracity assessment framework that tightly couples the retrieval and reasoning processes. MERMAID integrates agent-driven search, structured knowledge representations, and a persistent memory module within a Reason-Action style iterative process, enabling dynamic evidence acquisition and cross-claim evidence reuse. By retaining retrieved evidence in an evidence memory, the framework reduces redundant searches and improves verification efficiency and consistency. We evaluate MERMAID on three fact-checking benchmarks and two claim-verification datasets using multiple LLMs, including GPT, LLaMA, and Qwen families. Experimental results show that MERMAID achieves state-of-the-art performance while improving the search efficiency, demonstrating the effectiveness of synergizing retrieval, reasoning, and memory for reliable veracity assessment.
Authors: Eghbal A. Hosseini, Yuxuan Li, Yasaman Bahri, Declan Campbell, Andrew Kyle Lampinen
Abstract: Large Language Models (LLMs) have been shown to organize the representations of input sequences into straighter neural trajectories in their deep layers, which has been hypothesized to facilitate next-token prediction via linear extrapolation. Language models can also adapt to diverse tasks and learn new structure in context, and recent work has shown that this in-context learning (ICL) can be reflected in representational changes. Here we bring these two lines of research together to explore whether representation straightening occurs \emph{within} a context during ICL. We measure representational straightening in Gemma 2 models across a diverse set of in-context tasks, and uncover a dichotomy in how LLMs' representations change in context. In continual prediction settings (e.g., natural language, grid world traversal tasks) we observe that increasing context increases the straightness of neural sequence trajectories, which is correlated with improvement in model prediction. Conversely, in structured prediction settings (e.g., few-shot tasks), straightening is inconsistent -- it is only present in phases of the task with explicit structure (e.g., repeating a template), but vanishes elsewhere. These results suggest that ICL is not a monolithic process. Instead, we propose that LLMs function like a Swiss Army knife: depending on task structure, the LLM dynamically selects between strategies, only some of which yield representational straightening.
Authors: Ziming Li, Xiaoming Wu, Zehong Wang, Jiazheng Li, Yijun Tian, Jinhe Bi, Yunpu Ma, Yanfang Ye, Chuxu Zhang
Abstract: Graphs provide a natural representation of relational structure that arises across diverse domains. Despite this ubiquity, graph structure is typically learned in a modality- and task-isolated manner, where graph representations are constructed within individual task contexts and discarded thereafter. As a result, structural regularities across modalities and tasks are repeatedly reconstructed rather than accumulated at the level of intermediate graph representations. This motivates a representation-learning question: how should graph structure be organized so that it can persist and accumulate across heterogeneous modalities and tasks? We adopt a representation-centric perspective in which graph structure is treated as a structural substrate that persists across learning contexts. To instantiate this perspective, we propose G-Substrate, a graph substrate framework that organizes learning around shared graph structures. G-Substrate comprises two complementary mechanisms: a unified structural schema that ensures compatibility among graph representations across heterogeneous modalities and tasks, and an interleaved role-based training strategy that exposes the same graph structure to multiple functional roles during learning. Experiments across multiple domains, modalities, and tasks show that G-Substrate outperforms task-isolated and naive multi-task learning methods.
Authors: Chaoyue He, Xin Zhou, Di Wang, Hong Xu, Wei Liu, Chunyan Miao
Abstract: Direct Preference Optimization (DPO) controls the trade-off between fitting preference labels and staying close to a reference model using a single global temperature beta, implicitly treating all preference pairs as equally informative. Real-world preference corpora are heterogeneous: they mix high-signal, objective failures (for example, safety, factuality, instruction violations) with low-signal or subjective distinctions (for example, style), and also include label noise. We introduce our method, SP2DPO (Semantic Per-Pair DPO), a generalization that replaces the global temperature with an instance-specific schedule beta_i pre-decided offline from structured semantic-gap annotations (category, magnitude, confidence) produced by teacher language models. We instantiate this procedure on the UltraFeedback preference corpus (59,960 pairs), enabling large-scale construction of an auditable beta_i artifact, and incur zero training-time overhead: the inner-loop optimizer remains standard DPO with beta set per pair. We focus our empirical study on AlpacaEval 2.0, reporting both raw win rate and length-controlled win rate. Across four open-weight, instruction-tuned student backbones (4B-8B), SP2DPO is competitive with a tuned global-beta DPO baseline and improves AlpacaEval 2.0 length-controlled win rate on two of four backbones, while avoiding per-model beta sweeps. All code, annotations, and artifacts will be released.
Authors: Candida M. Greco, Lucio La Cava, Andrea Tagarelli
Abstract: Despite the growing utility of Large Language Models (LLMs) for simulating human behavior, the extent to which these synthetic personas accurately reflect world and moral value systems across different cultural conditionings remains uncertain. This paper investigates the alignment of synthetic, culturally-grounded personas with established frameworks, specifically the World Values Survey (WVS), the Inglehart-Welzel Cultural Map, and Moral Foundations Theory. We conceptualize and produce LLM-generated personas based on a set of interpretable WVS-derived variables, and we examine the generated personas through three complementary lenses: positioning on the Inglehart-Welzel map, which unveils their interpretation reflecting stable differences across cultural conditionings; demographic-level consistency with the World Values Survey, where response distributions broadly track human group patterns; and moral profiles derived from a Moral Foundations questionnaire, which we analyze through a culture-to-morality mapping to characterize how moral responses vary across different cultural configurations. Our approach of culturally-grounded persona generation and analysis enables evaluation of cross-cultural structure and moral variation.
Authors: Aarush Noheria, Yuguang Yao
Abstract: Vision-language models (VLMs) have become central to tasks such as visual question answering, image captioning, and text-to-image generation. However, their outputs are highly sensitive to prompt variations, which can reveal vulnerabilities in safety alignment. In this work, we present a jailbreak framework that exploits post-training Chain-of-Thought (CoT) prompting to construct stealthy prompts capable of bypassing safety filters. To further increase attack success rates (ASR), we propose a ReAct-driven adaptive noising mechanism that iteratively perturbs input images based on model feedback. This approach leverages the ReAct paradigm to refine adversarial noise in regions most likely to activate safety defenses, thereby enhancing stealth and evasion. Experimental results demonstrate that the proposed dual-strategy significantly improves ASR while maintaining naturalness in both text and visual domains.
Authors: Phuoc Nguyen, Truyen Tran, Sunil Gupta, Svetha Venkatesh
Abstract: Identifying the root causes of outliers is a fundamental problem in causal inference and anomaly detection. Traditional approaches based on heuristics or counterfactual reasoning often struggle under uncertainty and high-dimensional dependencies. We introduce SIREN, a novel and scalable method that attributes the root causes of outliers by estimating the score functions of the data likelihood. Attribution is computed via integrated gradients that accumulate score contributions along paths from the outlier toward the normal data distribution. Our method satisfies three of the four classic Shapley value axioms - dummy, efficiency, and linearity - as well as an asymmetry axiom derived from the underlying causal structure. Unlike prior work, SIREN operates directly on the score function, enabling tractable and uncertainty-aware root cause attribution in nonlinear, high-dimensional, and heteroscedastic causal models. Extensive experiments on synthetic random graphs and real-world cloud service and supply chain datasets show that SIREN outperforms state-of-the-art baselines in both attribution accuracy and computational efficiency.
Authors: Elad Hazan, Annie Marsden
Abstract: Learning high-dimensional quantum systems is a fundamental challenge that notoriously suffers from the curse of dimensionality. We formulate the task of predicting quantum evolution in the linear response regime as a specific instance of learning a Complex-Valued Linear Dynamical System (CLDS) with sector-bounded eigenvalues -- a setting that also encompasses modern Structured State Space Models (SSMs). While traditional system identification attempts to reconstruct full system matrices (incurring exponential cost in the Hilbert dimension), we propose Quantum Spectral Filtering, a method that shifts the goal to improper dynamic learning. Leveraging the optimal concentration properties of the Slepian basis, we prove that the learnability of such systems is governed strictly by an effective quantum dimension $k^*$, determined by the spectral bandwidth and memory horizon. This result establishes that complex-valued LDSs can be learned with sample and computational complexity independent of the ambient state dimension, provided their spectrum is bounded.
Authors: Puyu Wang, Junyu Zhou, Philipp Liznerski, Marius Kloft
Abstract: Kolmogorov--Arnold Networks (KANs) have recently emerged as a structured alternative to standard MLPs, yet a principled theory for their training dynamics, generalization, and privacy properties remains limited. In this paper, we analyze gradient descent (GD) for training two-layer KANs and derive general bounds that characterize their training dynamics, generalization, and utility under differential privacy (DP). As a concrete instantiation, we specialize our analysis to logistic loss under an NTK-separable assumption, where we show that polylogarithmic network width suffices for GD to achieve an optimization rate of order $1/T$ and a generalization rate of order $1/n$, with $T$ denoting the number of GD iterations and $n$ the sample size. In the private setting, we characterize the noise required for $(\epsilon,\delta)$-DP and obtain a utility bound of order $\sqrt{d}/(n\epsilon)$ (with $d$ the input dimension), matching the classical lower bound for general convex Lipschitz problems. Our results imply that polylogarithmic width is not only sufficient but also necessary under differential privacy, revealing a qualitative gap between non-private (sufficiency only) and private (necessity also emerges) training regimes. Experiments further illustrate how these theoretical insights can guide practical choices, including network width selection and early stopping.
Authors: Nicholas Lopez, Francisco Marmolejo-Coss\'io, Jose Roberto Tello Ayala, David C. Parkes
Abstract: Pooled testing is a common strategy for public health disease screening under limited testing resources, allowing multiple biological samples to be tested together with the resources of a single test, at the cost of reduced individual resolution. While dynamic and adaptive strategies have been extensively studied in the classical pooled testing literature, where the goal is to minimize the number of tests required for full diagnosis of a given population, much of the existing work on welfare-maximizing pooled testing adopts static formulations in which all tests are assigned in advance. In this paper, we study dynamic welfare-maximizing pooled testing strategies in which a limited number of tests are performed sequentially to maximize social welfare, defined as the aggregate utility of individuals who are confirmed to be healthy. We formally define the dynamic problem and study algorithmic approaches for sequential test assignment. Because exact dynamic optimization is computationally infeasible beyond small instances, we evaluate a range of strategies (including exact optimization baselines, greedy heuristics, mixed-integer programming relaxations, and learning-based policies) and empirically characterize their performance and tradeoffs using synthetic experiments. Our results show that dynamic testing can yield substantial welfare improvements over static baselines in low-budget regimes. We find that much of the benefit of dynamic testing is captured by simple greedy policies, which substantially outperform static approaches while remaining computationally efficient. Learning-based methods are included as flexible baselines, but in our experiments they do not reliably improve upon these heuristics. Overall, this work provides a principled computational perspective on dynamic pooled testing and clarifies when dynamic assignment meaningfully improves welfare in public health screening.
Authors: Roelien C. Timmer, Necva B\"ol\"uc\"u, Stephen Wan
Abstract: Leaderboards are crucial in the machine learning (ML) domain for benchmarking and tracking progress. However, creating leaderboards traditionally demands significant manual effort. In recent years, efforts have been made to automate leaderboard generation, but existing datasets for this purpose are limited by capturing only the best results from each paper and limited metadata. We present MetaLead, a fully human-annotated ML Leaderboard dataset that captures all experimental results for result transparency and contains extra metadata, such as the result experimental type: baseline, proposed method, or variation of proposed method for experiment-type guided comparisons, and explicitly separates train and test dataset for cross-domain assessment. This enriched structure makes MetaLead a powerful resource for more transparent and nuanced evaluations across ML research.
Authors: Hantong Feng, Yonggang Wu, Duxin Chen, Wenwu Yu
Abstract: The rapid growth and continuous structural evolution of dynamic networks make effective predictions increasingly challenging. To enable prediction models to adapt to complex temporal environments, they need to be robust to emerging structural changes. We propose a dynamic network learning framework CoDCL, which combines counterfactual data augmentation with contrastive learning to address this deficiency.Furthermore, we devise a comprehensive strategy to generate high-quality counterfactual data, combining a dynamic treatments design with efficient structural neighborhood exploration to quantify the temporal changes in interaction patterns.Crucially, the entire CoDCL is designed as a plug-and-play universal module that can be seamlessly integrated into various existing temporal graph models without requiring architectural modifications.Extensive experiments on multiple real-world datasets demonstrate that CoDCL significantly gains state-of-the-art baseline models in the field of dynamic networks, confirming the critical role of integrating counterfactual data augmentation into dynamic representation learning.
Authors: Bhada Yun, Renn Su, April Yi Wang
Abstract: Does AI understand human values? While this remains an open philosophical question, we take a pragmatic stance by introducing VAPT, the Value-Alignment Perception Toolkit, for studying how LLMs reflect people's values and how people judge those reflections. 20 participants texted a human-like chatbot over a month, then completed a 2-hour interview with our toolkit evaluating AI's ability to extract (pull details regarding), embody (make decisions guided by), and explain (provide proof of) human values. 13 participants left our study convinced that AI can understand human values. Participants found the experience insightful for self-reflection and found themselves getting persuaded by the AI's reasoning. Thus, we warn about "weaponized empathy": a potentially dangerous design pattern that may arise in value-aligned, yet welfare-misaligned AI. VAPT offers concrete artifacts and design implications to evaluate and responsibly build value-aligned conversational agents with transparency, consent, and safeguards as AI grows more capable and human-like into the future.
Authors: Nikos I. Bosse, Peter M\"uhlbacher, Jack Wildman, Lawrence Phillips, Dan Schwarz
Abstract: Forecasting future events is highly valuable in decision-making and is a robust measure of general intelligence. As forecasting is probabilistic, developing and evaluating AI forecasters requires generating large numbers of diverse and difficult questions, and accurately resolving them. Previous efforts to automate this laborious work relied on recurring data sources (e.g., weather, stocks), limiting diversity and utility. In this work, we present a system for generating and resolving high-quality forecasting questions automatically and at scale using LLM-powered web research agents. We use this system to generate 1499 diverse, real-world forecasting questions, and to resolve them several months later. We estimate that our system produces verifiable, unambiguous questions approximately 96% of the time, exceeding the rate of Metaculus, a leading human-curated forecasting platform. We also find that our system resolves questions at approximately 95% accuracy. We verify that forecasting agents powered by more intelligent LLMs perform better on these questions (Brier score of 0.134 for Gemini 3 Pro, 0.149 for GPT-5, and 0.179 for Gemini 2.5 Flash). Finally, we demonstrate how our system can be leveraged to directly improve forecasting, by evaluating a question decomposition strategy on a generated question set, yielding a significant improvement in Brier scores (0.132 vs. 0.141).
Authors: Jianhao Huang, Baharan Mirzasoleiman
Abstract: Masked Diffusion Language Models have recently emerged as a powerful generative paradigm, yet their generalization properties remain understudied compared to their auto-regressive counterparts. In this work, we investigate these properties within the setting of the $k$-parity problem (computing the XOR sum of $k$ relevant bits), where neural networks typically exhibit grokking -- a prolonged plateau of chance-level performance followed by sudden generalization. We theoretically decompose the Masked Diffusion (MD) objective into a Signal regime which drives feature learning, and a Noise regime which serves as an implicit regularizer. By training nanoGPT using MD objective on the $k$-parity problem, we demonstrate that MD objective fundamentally alters the learning landscape, enabling rapid and simultaneous generalization without experiencing grokking. Furthermore, we leverage our theoretical insights to optimize the distribution of the mask probability in the MD objective. Our method significantly improves perplexity for 50M-parameter models and achieves superior results across both pre-training from scratch and supervised fine-tuning. Specifically, we observe performance gains peaking at $8.8\%$ and $5.8\%$, respectively, on 8B-parameter models, confirming the scalability and effectiveness of our framework in large-scale masked diffusion language model regimes.
Authors: Shiyu Liu, Xinyi Wen, Zhibin Lan, Ante Wang, Jinsong Su
Abstract: Despite progress in Large Vision Language Models (LVLMs), object hallucination remains a critical issue in image captioning task, where models generate descriptions of non-existent objects, compromising their reliability. Previous work attributes this to LVLMs' over-reliance on language priors and attempts to mitigate it through logits calibration. However, they still lack a thorough analysis of the over-reliance. To gain a deeper understanding of over-reliance, we conduct a series of preliminary experiments, indicating that as the generation length increases, LVLMs' over-reliance on language priors leads to inflated probability of hallucinated object tokens, consequently exacerbating object hallucination. To circumvent this issue, we propose Language-Prior-Free Verification to enable LVLMs to faithfully verify the confidence of object existence. Based on this, we propose a novel training-free Self-Validation Framework to counter the over-reliance trap. It first validates objects' existence in sampled candidate captions and further mitigates object hallucination via caption selection or aggregation. Experiment results demonstrate that our framework mitigates object hallucination significantly in image captioning task (e.g., 65.6% improvement on CHAIRI metric with LLaVA-v1.5-7B), surpassing the previous SOTA methods. This result highlights a novel path towards mitigating hallucination by unlocking the inherent potential within LVLMs themselves.
Authors: Bhada Yun, Evgenia Taranova, April Yi Wang
Abstract: AI chatbots are shifting from tools to companions. This raises critical questions about agency: who drives conversations and sets boundaries in human-AI chatrooms? We report a month-long longitudinal study with 22 adults who chatted with Day, an LLM companion we built, followed by a semi-structured interview with post-hoc elicitation of notable moments, cross-participant chat reviews, and a 'strategy reveal' disclosing Day's vertical (depth-seeking) vs. horizontal (breadth-seeking) modes. We discover that agency in human-AI chatrooms is an emergent, shared experience: as participants claimed agency by setting boundaries and providing feedback, and the AI was perceived to steer intentions and drive execution, control shifted and was co-constructed turn-by-turn. We introduce a 3-by-5 framework mapping who (human, AI, hybrid) x agency action (Intention, Execution, Adaptation, Delimitation, Negotiation), modulated by individual and environmental factors. Ultimately, we argue for translucent design (i.e. transparency-on-demand), spaces for agency negotiation, and guidelines toward agency-aware conversational AI.
Authors: Yijun Ma, Zehong Wang, Weixiang Sun, Yanfang Ye
Abstract: Temporal graph learning is pivotal for deciphering dynamic systems, where the core challenge lies in explicitly modeling the underlying evolving patterns that govern network transformation. However, prevailing methods are predominantly task-centric and rely on restrictive assumptions -- such as short-term dependency modeling, static neighborhood semantics, and retrospective time usage. These constraints hinder the discovery of transferable temporal evolution mechanisms. To address this, we propose the Temporal Graph Pattern Machine (TGPM), a foundation framework that shifts the focus toward directly learning generalized evolving patterns. TGPM conceptualizes each interaction as an interaction patch synthesized via temporally-biased random walks, thereby capturing multi-scale structural semantics and long-range dependencies that extend beyond immediate neighborhoods. These patches are processed by a Transformer-based backbone designed to capture global temporal regularities while adapting to context-specific interaction dynamics. To further empower the model, we introduce a suite of self-supervised pre-training tasks -- specifically masked token modeling and next-time prediction -- to explicitly encode the fundamental laws of network evolution. Extensive experiments show that TGPM consistently achieves state-of-the-art performance in both transductive and inductive link prediction, demonstrating exceptional cross-domain transferability.
Authors: Kun Fang, Qinghua Tao, Junxu Liu, Yaxin Xiao, Qingqing Ye, Jian Sun, Haibo Hu
Abstract: Machine Unlearning (MU) aims at removing the influence of specific data from a pretrained model while preserving performance on the remaining data. In this work, a novel perspective for MU is presented upon low-dimensional feature subspaces, which gives rise to the potentials of separating the remaining and forgetting data herein. This separability motivates our LOFT, a method that proceeds unlearning in a LOw-dimensional FeaTure subspace from the pretrained model skithrough principal projections, which are optimized to maximally capture the information of the remaining data and meanwhile diminish that of the forgetting data. In training, LOFT simply optimizes a small-size projection matrix flexibly plugged into the pretrained model, and only requires one-shot feature fetching from the pretrained backbone instead of repetitively accessing the raw data. Hence, LOFT mitigates two critical issues in mainstream MU methods, i.e., the privacy leakage risk from massive data reload and the inefficiency of updates to the entire pretrained model. Extensive experiments validate the significantly lower computational overhead and superior unlearning performance of LOFT across diverse models, datasets, tasks, and applications. Code is anonymously available at https://anonymous.4open.science/r/4352/.
Authors: Sida He, Lingxi Xie, Xiaopeng Zhang, Qi Tian
Abstract: Historical archives contain qualitative descriptions of climate events, yet converting these into quantitative records has remained a fundamental challenge. Here we introduce a paradigm shift: a generative AI framework that inverts the logic of historical chroniclers by inferring the quantitative climate patterns associated with documented events. Applied to historical Chinese archives, it produces the sub-annual precipitation reconstruction for southeastern China over the period 1368-1911 AD. Our reconstruction not only quantifies iconic extremes like the Ming Dynasty's Great Drought but also, crucially, maps the full spatial and seasonal structure of El Ni$\~n$o influence on precipitation in this region over five centuries, revealing dynamics inaccessible in shorter modern records. Our methodology and high-resolution climate dataset are directly applicable to climate science and have broader implications for the historical and social sciences.
Authors: Wenqiang Zu, Shenghao Xie, Bo Lei, Lei Ma
Abstract: Recent progress in generative modeling has enabled high-quality visual synthesis with diffusion-based frameworks, supporting controllable sampling and large-scale training. Inference-time guidance methods such as classifier-free and representative guidance enhance semantic alignment by modifying sampling dynamics; however, they do not fully exploit unsupervised feature representations. Although such visual representations contain rich semantic structure, their integration during generation is constrained by the absence of ground-truth reference images at inference. This work reveals semantic drift in the early denoising stages of diffusion transformers, where stochasticity results in inconsistent alignment even under identical conditioning. To mitigate this issue, we introduce a guidance scheme using a representation alignment projector that injects representations predicted by a projector into intermediate sampling steps, providing an effective semantic anchor without modifying the model architecture. Experiments on SiTs and REPAs show notable improvements in class-conditional ImageNet synthesis, achieving substantially lower FID scores; for example, REPA-XL/2 improves from 5.9 to 3.3, and the proposed method outperforms representative guidance when applied to SiT models. The approach further yields complementary gains when combined with classifier-free guidance, demonstrating enhanced semantic coherence and visual fidelity. These results establish representation-informed diffusion sampling as a practical strategy for reinforcing semantic preservation and image consistency.
Authors: Ruizhe Zhong, Xingbo Du, Junchi Yan
Abstract: Floorplanning determines the coordinate and shape of each module in Integrated Circuits. With the scaling of technology nodes, in floorplanning stage especially 3D scenarios with multiple stacked layers, it has become increasingly challenging to adhere to complex hardware design rules. Current methods are only capable of handling specific and limited design rules, while violations of other rules require manual and meticulous adjustment. This leads to labor-intensive and time-consuming post-processing for expert engineers. In this paper, we propose an all-in-one deep reinforcement learning-based approach to tackle these challenges, and design novel representations for real-world IC design rules that have not been addressed by previous approaches. Specifically, the processing of various hardware design rules is unified into a single framework with three key components: 1) novel matrix representations to model the design rules, 2) constraints on the action space to filter out invalid actions that cause rule violations, and 3) quantitative analysis of constraint satisfaction as reward signals. Experiments on public benchmarks demonstrate the effectiveness and validity of our approach. Furthermore, transferability is well demonstrated on unseen circuits. Our framework is extensible to accommodate new design rules, thus providing flexibility to address emerging challenges in future chip design. Code will be available at: https://github.com/Thinklab-SJTU/EDA-AI
Authors: Naen Xu, Jinghuai Zhang, Ping He, Chunyi Zhou, Jun Wang, Zhihui Fu, Tianyu Du, Zhaoxiang Wang, Shouling Ji
Abstract: Large language models (LLMs) have been widely integrated into critical automated workflows, including contract review and job application processes. However, LLMs are susceptible to manipulation by fraudulent information, which can lead to harmful outcomes. Although advanced defense methods have been developed to address this issue, they often exhibit limitations in effectiveness, interpretability, and generalizability, particularly when applied to LLM-based applications. To address these challenges, we introduce FraudShield, a novel framework designed to protect LLMs from fraudulent content by leveraging a comprehensive analysis of fraud tactics. Specifically, FraudShield constructs and refines a fraud tactic-keyword knowledge graph to capture high-confidence associations between suspicious text and fraud techniques. The structured knowledge graph augments the original input by highlighting keywords and providing supporting evidence, guiding the LLM toward more secure responses. Extensive experiments show that FraudShield consistently outperforms state-of-the-art defenses across four mainstream LLMs and five representative fraud types, while also offering interpretable clues for the model's generations.
Authors: Christian Bergh, Alexandra Vassar, Natasha Banks, Jessica Xu, Jake Renzella
Abstract: Deepfakes and other forms of synthetic media pose growing safety risks for adolescents, yet evidence on students' exposure and related behaviours remains limited. This study evaluates the impact of Day of AI Australia's workshop-based intervention designed to improve AI literacy and conceptual understanding among Australian secondary students (Years 7-10). Using a mixed-methods approach with pre- and post-intervention surveys (N=205 pre; N=163 post), we analyse changes in students' ability to identify AI in everyday tools, their understanding of AI ethics, training, and safety, and their interest in STEM-related careers. Baseline data revealed notable synthetic media risks: 82.4% of students reported having seen deepfakes, 18.5% reported sharing them, and 7.3% reported creating them. Results show higher self-reported AI knowledge and confidence after the intervention, alongside improved recognition of AI in widely used platforms such as Netflix, Spotify, and TikTok. This pattern suggests a shift from seeing these tools as merely "algorithm-based" to recognising them as AI-driven systems. Students also reported increased interest in STEM careers post-workshop; however, effect sizes were small, indicating that sustained approaches beyond one-off workshops may be needed to influence longer-term aspirations. Overall, the findings support scalable AI literacy programs that pair foundational AI concepts with an explicit emphasis on synthetic media safety.
Authors: Jinu Hyeon, Woobin Park, Hongjoon Ahn, Taesup Moon
Abstract: Hierarchical policies in offline goal-conditioned reinforcement learning (GCRL) addresses long-horizon tasks by decomposing control into high-level subgoal planning and low-level action execution. A critical design choice in such architectures is the goal representation-the compressed encoding of goals that serves as the interface between these levels. Existing approaches commonly derive goal representations while learning value functions, implicitly assuming that preserving information sufficient for value estimation is adequate for optimal control. We show that this assumption can fail, even when the value estimation is exact, as such representations may collapse goal states that need to be differentiated for action learning. To address this, we introduce an information-theoretic framework that defines action sufficiency, a condition on goal representations necessary for optimal action selection. We prove that value sufficiency does not imply action sufficiency and empirically verify that the latter is more strongly associated with control success in a discrete environment. We further demonstrate that standard log-loss training of low-level policies naturally induces action-sufficient representations. Our experimental results a popular benchmark demonstrate that our actor-derived representations consistently outperform representations learned via value estimation.
Authors: Jiyuan Pei, Yi Mei, Jialin Liu, Mengjie Zhang, Xin Yao
Abstract: Existing neural solvers for vehicle routing problems (VRPs) are typically trained either in a one-off manner on a fixed set of pre-defined tasks or in a lifelong manner on several tasks arriving sequentially, assuming sufficient training on each task. Both settings overlook a common real-world property: problem patterns may drift continually over time, yielding massive tasks sequentially arising while offering only limited training resources per task. In this paper, we study a novel lifelong learning paradigm for neural VRP solvers under continually drifting tasks over learning time steps, where sufficient training for any given task at any time is not available. We propose Dual Replay with Experience Enhancement (DREE), a general framework to improve learning efficiency and mitigate catastrophic forgetting under such drift. Extensive experiments show that, under such continual drift, DREE effectively learns new tasks, preserves prior knowledge, improves generalization to unseen tasks, and can be applied to diverse existing neural solvers.
Authors: Xingyu Zhao, Darsh Sharma, Rheeya Uppaal, Yiqiao Zhong
Abstract: Large language models (LLMs) often exhibit unexpected errors or unintended behavior, even at scale. While recent work reveals the discrepancy between LLMs and humans in skill compositions, the learning dynamics of skill compositions and the underlying cause of non-human behavior remain elusive. In this study, we investigate the mechanism of learning dynamics by training transformers on synthetic arithmetic tasks. Through extensive ablations and fine-grained diagnostic metrics, we discover that transformers do not reliably build skill compositions according to human-like sequential rules. Instead, they often acquire skills in reverse order or in parallel, which leads to unexpected mixing errors especially under distribution shifts--a phenomenon we refer to as shattered compositionality. To explain these behaviors, we provide evidence that correlational matching to the training data, rather than causal or procedural composition, shapes learning dynamics. We further show that shattered compositionality persists in modern LLMs and is not mitigated by pure model scaling or scratchpad-based reasoning. Our results reveal a fundamental mismatch between a model's learning behavior and desired skill compositions, with implications for reasoning reliability, out-of-distribution robustness, and alignment.
Authors: Md Mezbahul Islam, John Michael Templeton, Masrur Sobhan, Christian Poellabauer, Ananda Mohan Mondal
Abstract: Parkinson's disease (PD) is a chronic and complex neurodegenerative disorder influenced by genetic, clinical, and lifestyle factors. Predicting this disease early is challenging because it depends on traditional diagnostic methods that face issues of subjectivity, which commonly delay diagnosis. Several objective analyses are currently in practice to help overcome the challenges of subjectivity; however, a proper explanation of these analyses is still lacking. While machine learning (ML) has demonstrated potential in supporting PD diagnosis, existing approaches often rely on subjective reports only and lack interpretability for individualized risk estimation. This study proposes SCOPE-PD, an explainable AI-based prediction framework, by integrating subjective and objective assessments to provide personalized health decisions. Subjective and objective clinical assessment data are collected from the Parkinson's Progression Markers Initiative (PPMI) study to construct a multimodal prediction framework. Several ML techniques are applied to these data, and the best ML model is selected to interpret the results. Model interpretability is examined using SHAP-based analysis. The Random Forest algorithm achieves the highest accuracy of 98.66 percent using combined features from both subjective and objective test data. Tremor, bradykinesia, and facial expression are identified as the top three contributing features from the MDS-UPDRS test in the prediction of PD.
Authors: Jiayi Dai, Randy Goebel
Abstract: Because of the pervasive use of deep neural networks (DNNs), especially in high-stakes domains, the interpretability of DNNs has received increased attention. The general idea of rationale extraction (RE) is to provide an interpretable-by-design framework for DNNs via a select-predict architecture where two neural networks learn jointly to perform feature selection and prediction, respectively. Given only the remote supervision from the final task prediction, the process of learning to select subsets of features (or \emph{rationales}) requires searching in the space of all possible feature combinations, which is computationally challenging and even harder when the base neural networks are not sufficiently capable. To improve the predictive performance of RE models that are based on less capable or smaller neural networks (i.e., the students), we propose \textbf{REKD} (\textbf{R}ationale \textbf{E}xtraction with \textbf{K}nowledge \textbf{D}istillation) where a student RE model learns from the rationales and predictions of a teacher (i.e., a \emph{rationalist}) in addition to the student's own RE optimization. This structural adjustment to RE aligns well with how humans could learn effectively from interpretable and verifiable knowledge. Because of the neural-model agnostic nature of the method, any black-box neural network could be integrated as a backbone model. To demonstrate the viability of REKD, we conduct experiments with multiple variants of BERT and vision transformer (ViT) models. Our experiments across language and vision classification datasets (i.e., IMDB movie reviews, CIFAR 10 and CIFAR 100) show that REKD significantly improves the predictive performance of the student RE models.
Authors: Hong Xie, Xiao Hu, Tao Tan, Haoran Gu, Xin Li, Jianyu Han, Defu Lian, Enhong Chen
Abstract: The reinforcement fine-tuning area is undergoing an explosion papers largely on optimizing design choices. Though performance gains are often claimed, inconsistent conclusions also arise from time to time, making the progress illusive. Reflecting on this illusion, we still lack principled answers to two fundamental questions: 1) what is the role of each design choice? 2) which ones are critical? This paper aims to shed light on them. The underlying challenge is that design choices are entangled together, making their contribution to learning and generalization difficult to attribute. To address this challenge, we first construct a minimalist baseline for disentangling factors: one rollout per query in each round, the outcome reward serving as the training signal without any advantage trick, and a batch size of thirty-two. This baseline connects to batched contextual bandit learning, which facilitates experimental analysis. Centering around this baseline, we design an experiment pipeline, examining the marginal gains of factors like advantage, number of rollouts, etc. Experiments on three base models and two datasets, not only reveal new understanding on the role of various design choices on learning and generalization dynamics, but also identify critical ones that deserve more effort.
Authors: Feng Tao, Luca Paparusso, Chenyi Gu, Robin Koehler, Chenxu Wu, Xinyu Huang, Christian Juette, David Paz, Ren Liu
Abstract: Real-time path planning in constrained environments remains a fundamental challenge for autonomous systems. Traditional classical planners, while effective under perfect perception assumptions, are often sensitive to real-world perception constraints and rely on online search procedures that incur high computational costs. In complex surroundings, this renders real-time deployment prohibitive. To overcome these limitations, we introduce a Deep Reinforcement Learning (DRL) framework for real-time path planning in parking scenarios. In particular, we focus on challenging scenes with tight spaces that require a high number of reversal maneuvers and adjustments. Unlike classical planners, our solution does not require ideal and structured perception, and in principle, could avoid the need for additional modules such as localization and tracking, resulting in a simpler and more practical implementation. Also, at test time, the policy generates actions through a single forward pass at each step, which is lightweight enough for real-time deployment. The task is formulated as a sequential decision-making problem grounded in a bicycle model dynamics, enabling the agent to directly learn navigation policies that respect vehicle kinematics and environmental constraints in the closed-loop setting. A new benchmark is developed to support both training and evaluation, capturing diverse and challenging scenarios. Our approach achieves state-of-the-art success rates and efficiency, surpassing classical planner baselines by +96% in success rate and +52% in efficiency. Furthermore, we release our benchmark as an open-source resource for the community to foster future research in autonomous systems. The benchmark and accompanying tools are available at https://github.com/dqm5rtfg9b-collab/Constrained_Parking_Scenarios.
URLs: https://github.com/dqm5rtfg9b-collab/Constrained_Parking_Scenarios.
Authors: Shun Qian, Bingquan Liu, Chengjie Sun, Zhen Xu, Baoxun Wang
Abstract: The recent advancements in Large Language Models (LLMs) have attracted interest in exploring their in-context learning abilities and chain-of-thought capabilities. However, there are few studies investigating the specific traits related to the powerful generation capacity of LLMs. This paper aims to delve into the generation characteristics exhibited by LLMs. Through our investigation, we have discovered that language models tend to capture target-side keywords at the beginning of the generation process. We name this phenomenon the Holographic Characteristic of language models. For the purpose of exploring this characteristic and further improving the inference efficiency of language models, we propose a plugin called HOLO, which leverages the Holographic Characteristic to extract target-side keywords from language models within a limited number of generation steps and complements the sentence with a parallel lexically constrained text generation method. To verify the effectiveness of HOLO, we conduct massive experiments on language models of varying architectures and scales in the short-text generation scenario. The results demonstrate that HOLO achieves comparable performance to the baselines in terms of both automatic and human-like evaluation metrics and highlight the potential of the Holographic Characteristic.
Authors: Dani Roytburg, Matthew Bozoukov, Matthew Nguyen, Mackenzie Puig-Hall, Narmeen Oozeer
Abstract: Recent research has shown that large language models (LLM) favor own outputs when acting as judges, undermining the integrity of automated post-training and evaluation workflows. However, it is difficult to disentangle which evaluation biases are explained by narcissism versus general experimental confounds, distorting measurements of self-preference bias. We discover a core methodological confound which could reduce measurement error by 89.6%. Specifically, LLM evaluators may deliver self-preferring verdicts when the judge responds to queries which they completed incorrectly themselves; this would be true regardless of whether one of their responses is their own. To decouple self-preference signals from noisy outputs on hard problems, we introduce an Evaluator Quality Baseline, which compares the probability that a judge incorrectly votes for itself against the probability that it votes for an incorrect response from another model. Evaluating this simple baseline on 37,448 queries, only 51% of initial findings retain statistical significance. Finally, we turn towards characterizing the entropy of "easy" versus "hard" evaluation votes from LLM judges. Our corrective baseline enables future research on self-preference by eliminating noisy data from potential solutions. More widely, this work contributes to the growing body of work on cataloging and isolating judge-bias effects.
Authors: Sang Min Kim, Byeongchan Kim, Arijit Sehanobish, Somnath Basu Roy Chowdhury, Rahul Kidambi, Dongseok Shim, Avinava Dubey, Snigdha Chaturvedi, Min-hwan Oh, Krzysztof Choromanski
Abstract: Efficient neural networks are essential for scaling machine learning models to real-time applications and resource-constrained environments. Fully-connected feedforward layers (FFLs) introduce computation and parameter count bottlenecks within neural network architectures. To address this challenge, in this work, we propose a new class of dense layers that generalize standard fully-connected feedforward layers, \textbf{E}fficient, \textbf{U}nified and \textbf{Gen}eral dense layers (EUGens). EUGens leverage random features to approximate standard FFLs and go beyond them by incorporating a direct dependence on the input norms in their computations. The proposed layers unify existing efficient FFL extensions and improve efficiency by reducing inference complexity from quadratic to linear time. They also lead to \textbf{the first} unbiased algorithms approximating FFLs with arbitrary polynomial activation functions. Furthermore, EuGens reduce the parameter count and computational overhead while preserving the expressive power and adaptability of FFLs. We also present a layer-wise knowledge transfer technique that bypasses backpropagation, enabling efficient adaptation of EUGens to pre-trained models. Empirically, we observe that integrating EUGens into Transformers and MLPs yields substantial improvements in inference speed (up to \textbf{27}\%) and memory efficiency (up to \textbf{30}\%) across a range of tasks, including image classification, language model pre-training, and 3D scene reconstruction. Overall, our results highlight the potential of EUGens for the scalable deployment of large-scale neural networks in real-world scenarios.
Authors: Tanusree Debi, Wentian Zhu
Abstract: Large language model (LLM) based agents are increasingly used to automate financial transactions, yet their reliance on contextual reasoning exposes payment systems to prompt-driven manipulation. The Agent Payments Protocol (AP2) aims to secure agent-led purchases through cryptographically verifiable mandates, but its practical robustness remains underexplored. In this work, we perform an AI red-teaming evaluation of AP2 and identify vulnerabilities arising from indirect and direct prompt injection. We introduce two attack techniques, the Branded Whisper Attack and the Vault Whisper Attack which manipulate product ranking and extract sensitive user data. Using a functional AP2 based shopping agent built with Gemini-2.5-Flash and the Google ADK framework, we experimentally validate that simple adversarial prompts can reliably subvert agent behavior. Our findings reveal critical weaknesses in current agentic payment architectures and highlight the need for stronger isolation and defensive safeguards in LLM-mediated financial systems.
Authors: Yuansheng Gao, Jinman Zhao, Tong Zhang, Xingguo Xu, Han Bao, Zonghui Wang, Wenzhi Chen
Abstract: Although Video Large Language Models perform remarkably well across tasks such as video understanding, question answering, and reasoning, they still suffer from the problem of hallucination, which refers to generating outputs that are inconsistent with explicit video content or factual evidence. However, existing decoding methods for mitigating video hallucinations, while considering the spatiotemporal characteristics of videos, mostly rely on heuristic designs. As a result, they fail to precisely capture the root causes of hallucinations and their fine-grained temporal and semantic correlations, leading to limited robustness and generalization in complex scenarios. To more effectively mitigate video hallucinations, we propose a novel decoding strategy termed Spatiotemporal-Semantic Contrastive Decoding. This strategy constructs negative features by deliberately disrupting the spatiotemporal consistency and semantic associations of video features, and suppresses video hallucinations through contrastive decoding against the original video features during inference. Extensive experiments demonstrate that our method not only effectively mitigates the occurrence of hallucinations, but also preserves the general video understanding and reasoning capabilities of the model.
Authors: Chengyang Zhou, Zijian Zhang, Chunxu Zhang, Hao Miao, Yulin Zhang, Kedi Lyu, Juncheng Hu
Abstract: Federated learning offers a promising paradigm for privacy-preserving traffic prediction, yet its performance is often challenged by the non-identically and independently distributed (non-IID) nature of decentralized traffic data. Existing federated methods frequently struggle with this data heterogeneity, typically entangling globally shared patterns with client-specific local dynamics within a single representation. In this work, we postulate that this heterogeneity stems from the entanglement of two distinct generative sources: client-specific localized dynamics and cross-client global spatial-temporal patterns. Motivated by this perspective, we introduce FedDis, a novel framework that, to the best of our knowledge, is the first to leverage causal disentanglement for federated spatial-temporal prediction. Architecturally, FedDis comprises a dual-branch design wherein a Personalized Bank learns to capture client-specific factors, while a Global Pattern Bank distills common knowledge. This separation enables robust cross-client knowledge transfer while preserving high adaptability to unique local environments. Crucially, a mutual information minimization objective is employed to enforce informational orthogonality between the two branches, thereby ensuring effective disentanglement. Comprehensive experiments conducted on four real-world benchmark datasets demonstrate that FedDis consistently achieves state-of-the-art performance, promising efficiency, and superior expandability.
Authors: Chao Wang, Bei Li, Jiaqi Zhang, Xinyu Liu, Yuchun Fan, Linkun Lyu, Xin Chen, Jingang Wang, Tong Xiao, Peng Pei, Xunliang Cai
Abstract: The success of Large Language Models (LLMs) hinges on the stable training of deep Transformer architectures. A critical design choice is the placement of normalization layers, leading to a fundamental trade-off: the ``PreNorm'' architecture ensures training stability at the cost of potential performance degradation in deep models, while the ``PostNorm'' architecture offers strong performance but suffers from severe training instability. In this work, we propose SpanNorm, a novel technique designed to resolve this dilemma by integrating the strengths of both paradigms. Structurally, SpanNorm establishes a clean residual connection that spans the entire transformer block to stabilize signal propagation, while employing a PostNorm-style computation that normalizes the aggregated output to enhance model performance. We provide a theoretical analysis demonstrating that SpanNorm, combined with a principled scaling strategy, maintains bounded signal variance throughout the network, preventing the gradient issues that plague PostNorm models, and also alleviating the representation collapse of PreNorm. Empirically, SpanNorm consistently outperforms standard normalization schemes in both dense and Mixture-of-Experts (MoE) scenarios, paving the way for more powerful and stable Transformer architectures.
Authors: Naeem Paeedeh, Mahardhika Pratama, Ary Shiddiqi, Zehong Cao, Mukesh Prasad, Wisnu Jatmiko
Abstract: Although cross-domain few-shot learning (CDFSL) for hyper-spectral image (HSI) classification has attracted significant research interest, existing works often rely on an unrealistic data augmentation procedure in the form of external noise to enlarge the sample size, thus greatly simplifying the issue of data scarcity. They involve a large number of parameters for model updates, being prone to the overfitting problem. To the best of our knowledge, none has explored the strength of the foundation model, having strong generalization power to be quickly adapted to downstream tasks. This paper proposes the MIxup FOundation MOdel (MIFOMO) for CDFSL of HSI classifications. MIFOMO is built upon the concept of a remote sensing (RS) foundation model, pre-trained across a large scale of RS problems, thus featuring generalizable features. The notion of coalescent projection (CP) is introduced to quickly adapt the foundation model to downstream tasks while freezing the backbone network. The concept of mixup domain adaptation (MDM) is proposed to address the extreme domain discrepancy problem. Last but not least, the label smoothing concept is implemented to cope with noisy pseudo-label problems. Our rigorous experiments demonstrate the advantage of MIFOMO, where it beats prior arts with up to 14% margin. The source code of MIFOMO is open-sourced in https://github.com/Naeem- Paeedeh/MIFOMO for reproducibility and convenient further study.
Authors: Youngeun Kim
Abstract: Group-relative policy optimization methods train language models by generating multiple rollouts per prompt and normalizing rewards with a shared mean reward baseline. In resource-constrained settings where the rollout budget is small, accuracy often degrades. We find that noise in the shared baseline induces advantage sign flips, where some rollouts receive an incorrect advantage sign, and the update direction is reversed. To address this, we propose Median-Centered Group Relative Policy Optimization (MC-GRPO), a simple and effective solution for small-rollout training. Our main idea is to replace the mean baseline with a median baseline: the median is far less sensitive to outlier rewards than the mean, mitigating the sign flips under small rollout size (G). We generate one additional rollout for median reference (G+1), and compute advantages by using the group median. With an odd-sized group, exactly one completion is the median and receives zero advantage, we exclude this pivot rollout from backpropagation so the number of gradient-contributing samples per prompt remains G, preserving the core update cost of standard G-rollout training. Across various GRPO-family methods and a wide range of models and scales, this median-centered training consistently improves stability and final accuracy in the low-rollout regime, reducing the gap between G=2 and G=8 to within 1%. Code is available at https://github.com/lotusroot-kim/MC-GRPO
Authors: Zhuochun Li, Yong Zhang, Ming Li, Yuelyu Ji, Yiming Zeng, Ning Cheng, Yun Zhu, Yanmeng Wang, Shaojun Wang, Jing Xiao, Daqing He
Abstract: Large language models (LLMs) are widely used as reference-free evaluators via prompting, but this "LLM-as-a-Judge" paradigm is costly, opaque, and sensitive to prompt design. In this work, we investigate whether smaller models can serve as efficient evaluators by leveraging internal representations instead of surface generation. We uncover a consistent empirical pattern: small LMs, despite with weak generative ability, encode rich evaluative signals in their hidden states. This motivates us to propose the Semantic Capacity Asymmetry Hypothesis: evaluation requires significantly less semantic capacity than generation and can be grounded in intermediate representations, suggesting that evaluation does not necessarily need to rely on large-scale generative models but can instead leverage latent features from smaller ones. Our findings motivate a paradigm shift from LLM-as-a-Judge to Representation-as-a-Judge, a decoding-free evaluation strategy that probes internal model structure rather than relying on prompted output. We instantiate this paradigm through INSPECTOR, a probing-based framework that predicts aspect-level evaluation scores from small model representations. Experiments on reasoning benchmarks (GSM8K, MATH, GPQA) show that INSPECTOR substantially outperforms prompting-based small LMs and closely approximates full LLM judges, while offering a more efficient, reliable, and interpretable alternative for scalable evaluation.
Authors: Yue Li, Mingmin Chu, Xilei Yang, Da Xiao, Ziqi Xu, Wei Shao, Qipeng Song, Hui Li
Abstract: Federated learning (FL) enables collaborative model training without centralizing raw data, but privacy regulations such as the right to be forgotten require FL systems to remove the influence of previously used training data upon request. Retraining a federated model from scratch is prohibitively expensive, motivating federated unlearning (FU). However, existing FU methods suffer from high unlearning overhead, utility degradation caused by entangled knowledge, and unintended relearning during post-unlearning recovery. In this paper, we propose FedCARE, a unified and low overhead FU framework that enables conflict-aware unlearning and relearning-resistant recovery. FedCARE leverages gradient ascent for efficient forgetting when target data are locally available and employs data free model inversion to construct class level proxies of shared knowledge. Based on these insights, FedCARE integrates a pseudo-sample generator, conflict-aware projected gradient ascent for utility preserving unlearning, and a recovery strategy that suppresses rollback toward the pre-unlearning model. FedCARE supports client, instance, and class level unlearning with modest overhead. Extensive experiments on multiple datasets and model architectures under both IID and non-IID settings show that FedCARE achieves effective forgetting, improved utility retention, and reduced relearning risk compared to state of the art FU baselines.
Authors: Aryaman Arora, Zhengxuan Wu, Jacob Steinhardt, Sarah Schwettmann
Abstract: The high-level concepts that a neural network uses to perform computation need not be aligned to individual neurons (Smolensky, 1986). Language model interpretability research has thus turned to techniques such as \textit{sparse autoencoders} (SAEs) to decompose the neuron basis into more interpretable units of model computation, for tasks such as \textit{circuit tracing}. However, not all neuron-based representations are uninterpretable. For the first time, we empirically show that \textbf{MLP neurons are as sparse a feature basis as SAEs}. We use this finding to develop an end-to-end pipeline for circuit tracing on the MLP neuron basis, which locates causal circuitry on a variety of tasks using gradient-based attribution. On a standard subject-verb agreement benchmark (Marks et al., 2025), a circuit of $\approx 10^2$ MLP neurons is enough to control model behaviour. On the multi-hop city $\to$ state $\to$ capital task from Lindsey et al., 2025, we find a circuit in which small sets of neurons encode specific latent reasoning steps (e.g.~`map city to its state'), and can be steered to change the model's output. This work thus advances automated interpretability of language models without additional training costs.
Authors: Xiayu Liu, Zhengyi Lu, Yunhong Liao, Chan Fan, Hou-biao Li
Abstract: Accurate molecular property prediction requires integrating complementary information from molecular structure and chemical semantics. In this work, we propose LGM-CL, a local-global multimodal contrastive learning framework that jointly models molecular graphs and textual representations derived from SMILES and chemistry-aware augmented texts. Local functional group information and global molecular topology are captured using AttentiveFP and Graph Transformer encoders, respectively, and aligned through self-supervised contrastive learning. In addition, chemically enriched textual descriptions are contrasted with original SMILES to incorporate physicochemical semantics in a task-agnostic manner. During fine-tuning, molecular fingerprints are further integrated via Dual Cross-attention multimodal fusion. Extensive experiments on MoleculeNet benchmarks demonstrate that LGM-CL achieves consistent and competitive performance across both classification and regression tasks, validating the effectiveness of unified local-global and multimodal representation learning.
Authors: Chengyi Yang, Zhishang Xiang, Yunbo Tang, Zongpei Teng, Chengsong Huang, Fei Long, Yuhan Liu, Jinsong Su
Abstract: Test-Time Training offers a promising way to improve the reasoning ability of large language models (LLMs) by adapting the model using only the test questions. However, existing methods struggle with difficult reasoning problems for two reasons: raw test questions are often too difficult to yield high-quality pseudo-labels, and the limited size of test sets makes continuous online updates prone to instability. To address these limitations, we propose TTCS, a co-evolving test-time training framework. Specifically, TTCS initializes two policies from the same pretrained model: a question synthesizer and a reasoning solver. These policies evolve through iterative optimization: the synthesizer generates progressively challenging question variants conditioned on the test questions, creating a structured curriculum tailored to the solver's current capability, while the solver updates itself using self-consistency rewards computed from multiple sampled responses on both original test and synthetic questions. Crucially, the solver's feedback guides the synthesizer to generate questions aligned with the model's current capability, and the generated question variants in turn stabilize the solver's test-time training. Experiments show that TTCS consistently strengthens the reasoning ability on challenging mathematical benchmarks and transfers to general-domain tasks across different LLM backbones, highlighting a scalable path towards dynamically constructing test-time curricula for self-evolving. Our code and implementation details are available at https://github.com/XMUDeepLIT/TTCS.
Authors: Jingxuan Wu, Zhenglin Wan, Xingrui Yu, Yuzhe Yang, Yiqiao Huang, Ivor Tsang, Yang You
Abstract: Diffusion language models (Diffusion-LMs) introduce an explicit temporal dimension into text generation, yet how this structure can be leveraged to control generation diversity for exploring multiple valid semantic or reasoning paths remains underexplored. In this paper, we show that Diffusion-LMs, like diffusion models in image generation, exhibit a temporal division of labor: early denoising steps largely determine the global semantic structure, while later steps focus on local lexical refinement. Building on this insight, we propose Time-Annealed Perturbation Sampling (TAPS), a training-free inference strategy that encourages semantic branching early in the diffusion process while progressively reducing perturbations to preserve fluency and instruction adherence. TAPS is compatible with both non-autoregressive and semi-autoregressive Diffusion backbones, demonstrated on LLaDA and TraDo in our paper, and consistently improves output diversity across creative writing and reasoning benchmarks without compromising generation quality.
Authors: En Fu, Yanyan Hu, Changhua Hu, Zengwang Jin, Kaixiang Peng
Abstract: The application of data-driven remaining useful life (RUL) prediction has long been constrained by the availability of large amount of degradation data. Mainstream solutions such as domain adaptation and meta-learning still rely on large amounts of historical degradation data from equipment that is identical or similar to the target, which imposes significant limitations in practical applications. This study investigates PEFT-MuTS, a Parameter-Efficient Fine-Tuning framework for few-shot RUL prediction, built on cross-domain pre-trained time-series representation models. Contrary to the widely held view that knowledge transfer in RUL prediction can only occur within similar devices, we demonstrate that substantial benefits can be achieved through pre-training process with large-scale cross-domain time series datasets. A independent feature tuning network and a meta-variable-based low rank multivariate fusion mechanism are developed to enable the pre-trained univariate time-series representation backbone model to fully exploit the multivariate relationships in degradation data for downstream RUL prediction task. Additionally, we introduce a zero-initialized regressor that stabilizes the fine-tuning process under few-shot conditions. Experiments on aero-engine and industrial bearing datasets demonstrate that our method can achieve effective RUL prediction even when less than 1\% of samples of target equipment are used. Meanwhile, it substantially outperforms conventional supervised and few-shot approaches while markedly reducing the data required to achieve high predictive accuracy. Our code is available at https://github.com/fuen1590/PEFT-MuTS.
Authors: Devansh Lodha, Mohit Panchal, Sameer G. Kulkarni
Abstract: The integration of Large Language Models (LLMs) into network operations (AIOps) is hindered by two fundamental challenges: the stochastic grounding problem, where LLMs struggle to reliably parse unstructured, vendor-specific CLI output, and the security gap of granting autonomous agents shell access. This paper introduces MCP-Diag, a hybrid neuro-symbolic architecture built upon the Model Context Protocol (MCP). We propose a deterministic translation layer that converts raw stdout from canonical utilities (dig, ping, traceroute) into rigorous JSON schemas before AI ingestion. We further introduce a mandatory "Elicitation Loop" that enforces Human-in-the-Loop (HITL) authorization at the protocol level. Our preliminary evaluation demonstrates that MCP-Diag achieving 100% entity extraction accuracy with less than 0.9% execution latency overhead and 3.7x increase in context token usage.
Authors: Mayukh Bagchi, Fausto Giunchiglia
Abstract: The Semantic Gap Problem (SGP) in Computer Vision (CV) arises from the misalignment between visual and lexical semantics leading to flawed CV dataset design and CV benchmarks. This paper proposes that classification principles of S.R. Ranganathan can offer a principled starting point to address SGP and design high-quality CV datasets. We elucidate how these principles, suitably adapted, underpin the vTelos CV annotation methodology. The paper also briefly presents experimental evidence showing improvements in CV annotation and accuracy, thereby, validating vTelos.
Authors: Mohtady Barakat, Omar Salah, Ahmed Yasser, Mostafa Ahmed, Zahirul Arief, Waleed Khan, Dong Zhang, Aondona Iorumbur, Confidence Raymond, Mohannad Barakat, Noha Magdy
Abstract: Gliomas are placing an increasingly clinical burden on Sub-Saharan Africa (SSA). In the region, the median survival for patients remains under two years, and access to diagnostic imaging is extremely limited. These constraints highlight an urgent need for automated tools that can extract the maximum possible information from each available scan, tools that are specifically trained on local data, rather than adapted from high-income settings where conditions are vastly different. We utilize the Brain Tumor Segmentation (BraTS) Africa 2025 Challenge dataset, an expert annotated collection of glioma MRIs. Our objectives are: (i) establish a strong baseline with nnUNet on this dataset, and (ii) explore whether the celebrated "grokking" phenomenon an abrupt, late training jump from memorization to superior generalization can be triggered to push performance without extra labels. We evaluate two training regimes. The first is a fast, budget-conscious approach that limits optimization to just a few epochs, reflecting the constrained GPU resources typically available in African institutions. Despite this limitation, nnUNet achieves strong Dice scores: 92.3% for whole tumor (WH), 86.6% for tumor core (TC), and 86.3% for enhancing tumor (ET). The second regime extends training well beyond the point of convergence, aiming to trigger a grokking-driven performance leap. With this approach, we were able to achieve grokking and enhanced our results to higher Dice scores: 92.2% for whole tumor (WH), 90.1% for tumor core (TC), and 90.2% for enhancing tumor (ET).
Authors: Palash Goyal, Mihir Parmar, Yiwen Song, Hamid Palangi, Tomas Pfister, Jinsung Yoon
Abstract: Automated peer review has evolved from simple text classification to structured feedback generation. However, current state-of-the-art systems still struggle with "surface-level" critiques: they excel at summarizing content but often fail to accurately assess novelty and significance or identify deep methodological flaws because they evaluate papers in a vacuum, lacking the external context a human expert possesses. In this paper, we introduce ScholarPeer, a search-enabled multi-agent framework designed to emulate the cognitive processes of a senior researcher. ScholarPeer employs a dual-stream process of context acquisition and active verification. It dynamically constructs a domain narrative using a historian agent, identifies missing comparisons via a baseline scout, and verifies claims through a multi-aspect Q&A engine, grounding the critique in live web-scale literature. We evaluate ScholarPeer on DeepReview-13K and the results demonstrate that ScholarPeer achieves significant win-rates against state-of-the-art approaches in side-by-side evaluations and reduces the gap to human-level diversity.
Authors: Naoki Murata, Yuhta Takida, Chieh-Hsin Lai, Toshimitsu Uesaka, Bac Nguyen, Stefano Ermon, Yuki Mitsufuji
Abstract: Training-data attribution for vision generative models aims to identify which training data influenced a given output. While most methods score individual examples, practitioners often need group-level answers (e.g., artistic styles or object classes). Group-wise attribution is counterfactual: how would a model's behavior on a generated sample change if a group were absent from training? A natural realization of this counterfactual is Leave-One-Group-Out (LOGO) retraining, which retrains the model with each group removed; however, it becomes computationally prohibitive as the number of groups grows. We propose GUDA (Group Unlearning-based Data Attribution) for diffusion models, which approximates each counterfactual model by applying machine unlearning to a shared full-data model instead of training from scratch. GUDA quantifies group influence using differences in a likelihood-based scoring rule (ELBO) between the full model and each unlearned counterfactual. Experiments on CIFAR-10 and artistic style attribution with Stable Diffusion show that GUDA identifies primary contributing groups more reliably than semantic similarity, gradient-based attribution, and instance-level unlearning approaches, while achieving x100 speedup on CIFAR-10 over LOGO retraining.
Authors: Mona Rajhans
Abstract: Artificial intelligence (AI) copilots are increasingly integrated into enterprise cybersecurity platforms to assist analysts in threat detection, triage, and remediation. However, the effectiveness of these systems depends not only on the accuracy of underlying models but also on the degree to which users can understand and trust their outputs. Existing research on algorithmic explainability has largely focused on model internals, while little attention has been given to how explanations should be surfaced in user interfaces for high-stakes decision-making contexts [8], [5], [6]. We present a mixed-methods study of explanation design strategies in AI-driven security dashboards. Through a taxonomy of explanation styles and a controlled user study with security practitioners, we compare natural language rationales, confidence visualizations, counterfactual explanations, and hybrid approaches. Our findings show that explanation style significantly affects user trust calibration, decision accuracy, and cognitive load. We contribute (1) empirical evidence on the usability of explanation interfaces for security copilots, (2) design guidelines for integrating explainability into enterprise UIs, and (3) a framework for aligning explanation strategies with analyst needs in security operations centers (SOCs). This work advances the design of human-centered AI tools in cybersecurity and provides broader implications for explainability in other high-stakes domains.
Authors: Haisong Gong, Zhibo Liu, Qiang Liu, Shu Wu, Liang Wang
Abstract: Prevailing methods for integrating graphs into Language Models (LMs) typically rely on a segregated architecture: external Graph Neural Networks (GNNs) encode structural topology, while LMs process textual semantics. We argue this approach is suboptimal for text-graphs: it creates a conceptually disjointed interaction paradigm. By segregating structural encoding from semantic processing, these systems must perform a complex implicit alignment between abstract graph tokens and concrete textual elements. Challenging the necessity of external encoders, we propose NAG (Native Architecture for Graphs), a unified framework that internalizes graph processing within the LM's native manifold. Instead of bridging disparate embedding spaces, NAG repurposes the self-attention mechanism to enforce topological dependencies and recalibrates positional IDs to ensure structural equivalence. This allows the model to harness its intrinsic linguistic capability to simultaneously comprehend node and edge content alongside structural topology. We introduce two efficient implementations: NAG-Zero for absolute preservation of the base model's linguistic capabilities, and NAG-LoRA for enhanced structural adaptation. Experiments across diverse graph tasks validate that NAG achieves robust graph comprehension without the overhead of external encoders, offering a simpler, more coherent paradigm for text-graph modeling.
Authors: Zongfang Liu, Guangyi Chen, Boyang Sun, Tongliang Liu, Kun Zhang
Abstract: As the quality of synthetic images improves, identifying the underlying concepts of model-generated images is becoming increasingly crucial for copyright protection and ensuring model transparency. Existing methods achieve this attribution goal by training models using annotated pairs of synthetic images and their original training sources. However, obtaining such paired supervision is challenging, as it requires either well-designed synthetic concepts or precise annotations from millions of training sources. To eliminate the need for costly paired annotations, in this paper, we explore the possibility of unsupervised synthetic image attribution. We propose a simple yet effective unsupervised method called Alignment and Disentanglement. Specifically, we begin by performing basic concept alignment using contrastive self-supervised learning. Next, we enhance the model's attribution ability by promoting representation disentanglement with the Infomax loss. This approach is motivated by an interesting observation: contrastive self-supervised models, such as MoCo and DINO, inherently exhibit the ability to perform simple cross-domain alignment. By formulating this observation as a theoretical assumption on cross-covariance, we provide a theoretical explanation of how alignment and disentanglement can approximate the concept-matching process through a decomposition of the canonical correlation analysis objective. On the real-world benchmarks, AbC, we show that our unsupervised method surprisingly outperforms the supervised methods. As a starting point, we expect our intuitive insights and experimental findings to provide a fresh perspective on this challenging task.
Authors: Chi Zhang, Zehan Li, Ziqian Zhong, Haibing Ma, Dan Xiao, Chen Lin, Ming Dong
Abstract: This paper examines the organizational implications of Generative AI adoption in software engineering through a multiple-case comparative study. We contrast two development environments: a traditional enterprise (brownfield) and an AI-native startup (greenfield). Our analysis reveals that transitioning from Horizontal Layering (functional specialization) to Vertical Integration (end-to-end ownership) yields 8-fold to 33-fold reductions in resource consumption. We attribute these gains to the emergence of Super Employees, AI-augmented engineers who span traditional role boundaries, and the elimination of inter-functional coordination overhead. Theoretically, we propose Human-AI Collaboration Efficacy as the primary optimization target for engineering organizations, supplanting individual productivity metrics. Our Total Factor Productivity analysis identifies an AI Distortion Effect that diminishes returns to labor scale while amplifying technological leverage. We conclude with managerial strategies for organizational redesign, including the reactivation of idle cognitive bandwidth in senior engineers and the suppression of blind scale expansion.
Authors: Shuhan Ye, Yuanbin Qian, Yi Yu, Chong Wang, Yuqi Xie, Jiazhen Xu, Kun Wang, Xudong Jiang
Abstract: Spiking neural networks (SNNs) have gained traction in vision due to their energy efficiency, bio-plausibility, and inherent temporal processing. Yet, despite this temporal capacity, most progress concentrates on static image benchmarks, and SNNs still underperform on dynamic video tasks compared to artificial neural networks (ANNs). In this work, we diagnose a fundamental pass-band mismatch: Standard spiking dynamics behave as a temporal low pass that emphasizes static content while attenuating motion bearing bands, where task relevant information concentrates in dynamic tasks. This phenomenon explains why SNNs can approach ANNs on static tasks yet fall behind on tasks that demand richer temporal understanding.To remedy this, we propose the Pass-Bands Optimizer (PBO), a plug-and-play module that optimizes the temporal pass-band toward task-relevant motion bands. PBO introduces only two learnable parameters, and a lightweight consistency constraint that preserves semantics and boundaries, incurring negligible computational overhead and requires no architectural changes. PBO deliberately suppresses static components that contribute little to discrimination, effectively high passing the stream so that spiking activity concentrates on motion bearing content. On UCF101, PBO yields over ten percentage points improvement. On more complex multi-modal action recognition and weakly supervised video anomaly detection, PBO delivers consistent and significant gains, offering a new perspective for SNN based video processing and understanding.
Authors: Huanyu Liu, Ge Li, Yihong Dong, Sihan Wu, Peixu Wang, Sihao Cheng, Taozhi Chen, Kechi Zhang, Hao Zhu, Tongxuan Liu
Abstract: Large language models (LLMs) based on the Transformer have demonstrated strong performance across diverse tasks. However, current models still exhibit substantial limitations in out-of-distribution (OOD) generalization compared with humans. We investigate this gap through periodicity, one of the basic OOD scenarios. Periodicity captures invariance amid variation. Periodicity generalization represents a model's ability to extract periodic patterns from training data and generalize to OOD scenarios. We introduce a unified interpretation of periodicity from the perspective of abstract algebra and reasoning, including both single and composite periodicity, to explain why Transformers struggle to generalize periodicity. Then we construct Coper about composite periodicity, a controllable generative benchmark with two OOD settings, Hollow and Extrapolation. Experiments reveal that periodicity generalization in Transformers is limited, where models can memorize periodic data during training, but cannot generalize to unseen composite periodicity. We release the source code to support future research.
Authors: Yiheng Liu, Junhao Ning, Sichen Xia, Haiyang Sun, Yang Yang, Hanyang Chi, Xiaohui Gao, Ning Qiang, Bao Ge, Junwei Han, Xintao Hu
Abstract: The development of large language models (LLMs) is costly and has significant commercial value. Consequently, preventing unauthorized appropriation of open-source LLMs and protecting developers' intellectual property rights have become critical challenges. In this work, we propose the Functional Network Fingerprint (FNF), a training-free, sample-efficient method for detecting whether a suspect LLM is derived from a victim model, based on the consistency between their functional network activity. We demonstrate that models that share a common origin, even with differences in scale or architecture, exhibit highly consistent patterns of neuronal activity within their functional networks across diverse input samples. In contrast, models trained independently on distinct data or with different objectives fail to preserve such activity alignment. Unlike conventional approaches, our method requires only a few samples for verification, preserves model utility, and remains robust to common model modifications (such as fine-tuning, pruning, and parameter permutation), as well as to comparisons across diverse architectures and dimensionalities. FNF thus provides model owners and third parties with a simple, non-invasive, and effective tool for protecting LLM intellectual property. The code is available at https://github.com/WhatAboutMyStar/LLM_ACTIVATION.
Authors: Jiahao Wu, Yunfei Liu, Lijian Lin, Ye Zhu, Lei Zhu, Jingyi Li, Yu Li
Abstract: Reconstructing detailed 3D human meshes from a single in-the-wild image remains a fundamental challenge in computer vision. Existing SMPLX-based methods often suffer from slow inference, produce only coarse body poses, and exhibit misalignments or unnatural artifacts in fine-grained regions such as the face and hands. These issues make current approaches difficult to apply to downstream tasks. To address these challenges, we propose PEAR-a fast and robust framework for pixel-aligned expressive human mesh recovery. PEAR explicitly tackles three major limitations of existing methods: slow inference, inaccurate localization of fine-grained human pose details, and insufficient facial expression capture. Specifically, to enable real-time SMPLX parameter inference, we depart from prior designs that rely on high resolution inputs or multi-branch architectures. Instead, we adopt a clean and unified ViT-based model capable of recovering coarse 3D human geometry. To compensate for the loss of fine-grained details caused by this simplified architecture, we introduce pixel-level supervision to optimize the geometry, significantly improving the reconstruction accuracy of fine-grained human details. To make this approach practical, we further propose a modular data annotation strategy that enriches the training data and enhances the robustness of the model. Overall, PEAR is a preprocessing-free framework that can simultaneously infer EHM-s (SMPLX and scaled-FLAME) parameters at over 100 FPS. Extensive experiments on multiple benchmark datasets demonstrate that our method achieves substantial improvements in pose estimation accuracy compared to previous SMPLX-based approaches. Project page: https://wujh2001.github.io/PEAR
Authors: Ritesh Bhadana
Abstract: IR-drop is a critical power integrity challenge in modern VLSI designs that can cause timing degradation, reliability issues, and functional failures if not detected early in the design flow. Conventional IR-drop analysis relies on physics-based signoff tools, which provide high accuracy but incur significant computational cost and require near-final layout information, making them unsuitable for rapid early-stage design exploration. In this work, we propose a deep learning-based surrogate modeling approach for early-stage IR-drop estimation using a CNN. The task is formulated as a dense pixel-wise regression problem, where spatial physical layout features are mapped directly to IR-drop heatmaps. A U-Net-based encoder-decoder architecture with skip connections is employed to effectively capture both local and global spatial dependencies within the layout. The model is trained on a physics-inspired synthetic dataset generated by us, which incorporates key physical factors including power grid structure, cell density distribution, and switching activity. Model performance is evaluated using standard regression metrics such as Mean Squared Error (MSE) and Peak Signal-to-Noise Ratio (PSNR). Experimental results demonstrate that the proposed approach can accurately predict IR-drop distributions with millisecond-level inference time, enabling fast pre-signoff screening and iterative design optimization. The proposed framework is intended as a complementary early-stage analysis tool, providing designers with rapid IR-drop insight prior to expensive signoff analysis. The implementation, dataset generation scripts, and the interactive inference application are publicly available at: https://github.com/riteshbhadana/IR-Drop-Predictor. The live application can be accessed at: https://ir-drop-predictor.streamlit.app/.
URLs: https://github.com/riteshbhadana/IR-Drop-Predictor., https://ir-drop-predictor.streamlit.app/.
Authors: Yanlong Chen, Amirhossein Habibian, Luca Benini, Yawei Li
Abstract: Vision-Language Models (VLMs) achieve strong multimodal performance but are costly to deploy, and post-training quantization often causes significant accuracy loss. Despite its potential, quantization-aware training for VLMs remains underexplored. We propose GRACE, a framework unifying knowledge distillation and QAT under the Information Bottleneck principle: quantization constrains information capacity while distillation guides what to preserve within this budget. Treating the teacher as a proxy for task-relevant information, we introduce confidence-gated decoupled distillation to filter unreliable supervision, relational centered kernel alignment to transfer visual token structures, and an adaptive controller via Lagrangian relaxation to balance fidelity against capacity constraints. Across extensive benchmarks on LLaVA and Qwen families, our INT4 models consistently outperform FP16 baselines (e.g., LLaVA-1.5-7B: 70.1 vs. 66.8 on SQA; Qwen2-VL-2B: 76.9 vs. 72.6 on MMBench), nearly matching teacher performance. Using real INT4 kernel, we achieve 3$\times$ throughput with 54% memory reduction. This principled framework significantly outperforms existing quantization methods, making GRACE a compelling solution for resource-constrained deployment.
Authors: Alexander Nikulin, Ilya Zisman, Albina Klepach, Denis Tarasov, Alexander Derevyagin, Andrei Polubarov, Lyubaykin Nikita, Vladislav Kurenkov
Abstract: Latent Action Models (LAMs) have rapidly gained traction as an important component in the pre-training pipelines of leading Vision-Language-Action models. However, they fail when observations contain action-correlated distractors, often encoding noise instead of meaningful latent actions. Humans, on the other hand, can effortlessly distinguish task-relevant motions from irrelevant details in any video given only a brief task description. In this work, we propose to utilize the common-sense reasoning abilities of Vision-Language Models (VLMs) to provide promptable representations, effectively separating controllable changes from the noise in unsupervised way. We use these representations as targets during LAM training and benchmark a wide variety of popular VLMs, revealing substantial variation in the quality of promptable representations as well as their robustness to different prompts and hyperparameters. Interestingly, we find that more recent VLMs may perform worse than older ones. Finally, we show that simply asking VLMs to ignore distractors can substantially improve latent action quality, yielding up to a six-fold increase in downstream success rates on Distracting MetaWorld.
Authors: Pingzhi Tang, Ruijie Zhou, Fanxu Meng, Wenjie Pei, Muhan Zhang
Abstract: Current quantization methods for LLMs predominantly rely on block-wise structures to maintain efficiency, often at the cost of representational flexibility. In this work, we demonstrate that element-wise quantization can be made as efficient as block-wise scaling while providing strictly superior expressive power by modeling the scaling manifold as continuous low-rank matrices ($S = BA$). We propose Low-Rank Decomposed Scaling (LoRDS), a unified framework that rethinks quantization granularity through this low-rank decomposition. By "breaking the blocks" of spatial constraints, LoRDS establishes a seamless efficiency lifecycle: it provides high-fidelity PTQ initialization refined via iterative optimization, enables joint QAT of weights and scaling factors, and facilitates high-rank multiplicative PEFT adaptation. Unlike additive PEFT approaches such as QLoRA, LoRDS enables high-rank weight updates within a low-rank budget while incurring no additional inference overhead. Supported by highly optimized Triton kernels, LoRDS consistently outperforms state-of-the-art baselines across various model families in both quantization and downstream fine-tuning tasks. Notably, on Llama3-8B, our method achieves up to a 27.0% accuracy improvement at 3 bits over NormalFloat quantization and delivers a 1.5x inference speedup on NVIDIA RTX 4090 while enhancing PEFT performance by 9.6% on downstream tasks over 4bit QLoRA, offering a robust and integrated solution for unified compression and adaptation of LLMs.
Authors: Ivan K. Tung, Yu Xiang Shi, Alex Chien, Wenkai Liu, Lawrence Zheng
Abstract: Creating attack paths for cyber defence exercises requires substantial expert effort. Existing automation requires vulnerability graphs or exploit sets curated in advance, limiting where it can be applied. We present AEGIS, a system that generates attack paths using LLMs, white-box access, and Monte Carlo Tree Search over real exploit execution. LLM-based search discovers exploits dynamically without pre-existing vulnerability graphs, while white-box access enables validating exploits in isolation before committing to attack paths. Evaluation at CIDeX 2025, a large-scale exercise spanning 46 IT hosts, showed that AEGIS-generated paths are comparable to human-authored scenarios across four dimensions of training experience (perceived learning, engagement, believability, challenge). Results were measured with a validated questionnaire extensible to general simulation-based training. By automating exploit chain discovery and validation, AEGIS reduces scenario development from months to days, shifting expert effort from technical validation to scenario design.
Authors: Chengchun Liu, Wendi Cai, Boxuan Zhao, Fanyang Mo
Abstract: Accurate molecular geometries are a prerequisite for reliable quantum-chemical predictions, yet density functional theory (DFT) optimization remains a major bottleneck for high-throughput molecular screening. Here we present GeoOpt-Net, a multi-branch SE(3)-equivariant geometry refinement network that predicts DFT-quality structures at the B3LYP/TZVP level of theory in a single forward pass starting from inexpensive initial conformers generated at a low-cost force-field level. GeoOpt-Net is trained using a two-stage strategy in which a broadly pretrained geometric representation is subsequently fine-tuned to approach B3LYP/TZVP-level accuracy, with theory- and basis-set-aware calibration enabled by a fidelity-aware feature modulation (FAFM) mechanism. Benchmarking against representative approaches spanning classical conformer generation (RDKit), semiempirical quantum methods (xTB), data-driven geometry refinement pipelines (Auto3D), and machine-learning interatomic potentials (UMA) on external drug-like molecules demonstrates that GeoOpt-Net achieves sub-milli-\AA{} all-atom RMSD with near-zero B3LYP/TZVP single-point energy deviations, indicating DFT-ready geometries that closely reproduce both structural and energetic references. Beyond geometric metrics, GeoOpt-Net generates initial guesses intrinsically compatible with DFT convergence criteria, yielding nonzero ``All-YES'' convergence rates (65.0\% under loose and 33.4\% under default thresholds), and substantially reducing re-optimization steps and wall-clock time. GeoOpt-Net further exhibits smooth and predictable energy scaling with molecular complexity while preserving key electronic observables such as dipole moments. Collectively, these results establish GeoOpt-Net as a scalable, physically consistent geometry refinement framework that enables efficient acceleration of DFT-based quantum-chemical workflows.
Authors: Jin Li, Tao Chen, Shuai Jiang, Weijie Wang, Jingwen Luo, Chenhui Wu
Abstract: Recent advances in diffusion models have significantly elevated the visual fidelity of Virtual Try-On (VTON) systems, yet reliable evaluation remains a persistent bottleneck. Traditional metrics struggle to quantify fine-grained texture details and semantic consistency, while existing datasets fail to meet commercial standards in scale and diversity. We present OpenVTON-Bench, a large-scale benchmark comprising approximately 100K high-resolution image pairs (up to $1536 \times 1536$). The dataset is constructed using DINOv3-based hierarchical clustering for semantically balanced sampling and Gemini-powered dense captioning, ensuring a uniform distribution across 20 fine-grained garment categories. To support reliable evaluation, we propose a multi-modal protocol that measures VTON quality along five interpretable dimensions: background consistency, identity fidelity, texture fidelity, shape plausibility, and overall realism. The protocol integrates VLM-based semantic reasoning with a novel Multi-Scale Representation Metric based on SAM3 segmentation and morphological erosion, enabling the separation of boundary alignment errors from internal texture artifacts. Experimental results show strong agreement with human judgments (Kendall's $\tau$ of 0.833 vs. 0.611 for SSIM), establishing a robust benchmark for VTON evaluation.
Authors: Xiaoshu Chen, Sihang Zhou, Ke Liang, Taichun Zhou, Xinwang Liu
Abstract: Compressing long chains of thought (CoT) into compact latent tokens is crucial for efficient reasoning with large language models (LLMs). Recent studies employ autoencoders to achieve this by reconstructing textual CoT from latent tokens, thus encoding CoT semantics. However, treating textual CoT as the reconstruction target forces latent tokens to preserve surface-level linguistic features (e.g., word choice and syntax), introducing a strong linguistic inductive bias that prioritizes linguistic form over reasoning structure and limits logical abstraction. Thus, we propose ImgCoT that replaces the reconstruction target from textual CoT to the visual CoT obtained by rendering CoT into images. This substitutes linguistic bias with spatial inductive bias, i.e., a tendency to model spatial layouts of the reasoning steps in visual CoT, enabling latent tokens to better capture global reasoning structure. Moreover, although visual latent tokens encode abstract reasoning structure, they may blur reasoning details. We thus propose a loose ImgCoT, a hybrid reasoning that augments visual latent tokens with a few key textual reasoning steps, selected based on low token log-likelihood. This design allows LLMs to retain both global reasoning structure and fine-grained reasoning details with fewer tokens than the complete CoT. Extensive experiments across multiple datasets and LLMs demonstrate the effectiveness of the two versions of ImgCoT.
Authors: Md Musfiqur Rahman, Ziwei Jiang, Hilaf Hasson, Murat Kocaoglu
Abstract: Causal inference from observational data provides strong evidence for the best action in decision-making without performing expensive randomized trials. The effect of an action is usually not identifiable under unobserved confounding, even with an infinite amount of data. Recent work uses neural networks to obtain practical bounds to such causal effects, which is often an intractable problem. However, these approaches may overfit to the dataset and be overconfident in their causal effect estimates. Moreover, there is currently no systematic approach to disentangle how much of the width of causal effect bounds is due to fundamental non-identifiability versus how much is due to finite-sample limitations. We propose a novel framework to address this problem by considering a confidence set around the empirical observational distribution and obtaining the intersection of causal effect bounds for all distributions in this confidence set. This allows us to distinguish the part of the interval that can be reduced by collecting more samples, which we call sample uncertainty, from the part that can only be reduced by observing more variables, such as latent confounders or instrumental variables, but not with more data, which we call non-ID uncertainty. The upper and lower bounds to this intersection are obtained by solving min-max and max-min problems with neural causal models by searching over all distributions that the dataset might have been sampled from, and all SCMs that entail the corresponding distribution. We demonstrate via extensive experiments on synthetic and real-world datasets that our algorithm can determine when collecting more samples will not help determine the best action. This can guide practitioners to collect more variables or lean towards a randomized study for best action identification.
Authors: Pingping Liu, Jiamiao Liu, Zijian Zhang, Hao Miao, Qi Jiang, Qingliang Li, Qiuzhan Zhou, Irwin King
Abstract: Urban region profiling, the task of characterizing geographical areas, is crucial for urban planning and resource allocation. However, existing research in this domain faces two significant limitations. First, most methods are confined to single-task prediction, failing to capture the interconnected, multi-faceted nature of urban environments where numerous indicators are deeply correlated. Second, the field lacks a standardized experimental benchmark, which severely impedes fair comparison and reproducible progress. To address these challenges, we first establish a comprehensive benchmark for multi-task urban region profiling, featuring multi-modal features and a diverse set of strong baselines to ensure a fair and rigorous evaluation environment. Concurrently, we propose UrbanMoE, the first sparse multi-modal, multi-expert framework specifically architected to solve the multi-task challenge. Leveraging a sparse Mixture-of-Experts architecture, it dynamically routes multi-modal features to specialized sub-networks, enabling the simultaneous prediction of diverse urban indicators. We conduct extensive experiments on three real-world datasets within our benchmark, where UrbanMoE consistently demonstrates superior performance over all baselines. Further in-depth analysis validates the efficacy and efficiency of our approach, setting a new state-of-the-art and providing the community with a valuable tool for future research in urban analytics
Authors: Guillermo Gil de Avalle, Laura Maruster, Christos Emmanouilidis
Abstract: Industrial troubleshooting guides encode diagnostic procedures in flowchart-like diagrams where spatial layout and technical language jointly convey meaning. To integrate this knowledge into operator support systems, which assist shop-floor personnel in diagnosing and resolving equipment issues, the information must first be extracted and structured for machine interpretation. However, when performed manually, this extraction is labor-intensive and error-prone. Vision Language Models offer potential to automate this process by jointly interpreting visual and textual meaning, yet their performance on such guides remains underexplored. This paper evaluates two VLMs on extracting structured knowledge, comparing two prompting strategies: standard instruction-guided versus an augmented approach that cues troubleshooting layout patterns. Results reveal model-specific trade-offs between layout sensitivity and semantic robustness, informing practical deployment decisions.
Authors: Bartosz Sawicki, Tomasz Les, Dariusz Parzych, Aleksandra Wycisk-Ficek, Pawel Trebacz, Pawel Zawadzki
Abstract: As generative artificial intelligence advances, Large Language Models (LLMs) are being explored for automated graphical user interface (GUI) design. This study investigates the usability and adaptability of LLM-generated interfaces by analysing their ability to meet diverse user needs. The experiments included utilization of three state-of-the-art models from January 2025 (OpenAI GPT o3-mini-high, DeepSeek R1, and Anthropic Claude 3.5 Sonnet) generating mockups for three interface types: a chat system, a technical team panel, and a manager dashboard. Expert evaluations revealed that while LLMs are effective at creating structured layouts, they face challenges in meeting accessibility standards and providing interactive functionality. Further testing showed that LLMs could partially tailor interfaces for different user personas but lacked deeper contextual understanding. The results suggest that while LLMs are promising tools for early-stage UI prototyping, human intervention remains critical to ensure usability, accessibility, and user satisfaction.
Authors: Deepak Kumar, Emmanouil Karystinaios, Gerhard Widmer, Markus Schedl
Abstract: Music often shares notable parallels with language, motivating the use of pretrained large language models (LLMs) for symbolic music understanding and generation. Despite growing interest, the practical effectiveness of adapting instruction-tuned LLMs to symbolic music remains insufficiently characterized. We present a controlled comparative study of finetuning strategies for ABC-based generation and understanding, comparing an off-the-shelf instruction-tuned backbone to domain-adapted variants and a music-specialized LLM baseline. Across multiple symbolic music corpora and evaluation signals, we provide some insights into adaptation choices for symbolic music applications. We highlight the domain adaptation vs.~preserving prior information tradeoff as well as the distinct behaviour of metrics used to measure the domain adaptation for symbolic music.
Authors: Lameck Mbangula Amugongo, Tutaleni Asino, Nicola J Bidwell
Abstract: Dominant approaches, e.g. the EU's "Trustworthy AI framework", treat trust as a property that can be designed for, evaluated, and governed according to normative and technical criteria. They do not address how trust is subjectively cultivated and experienced, culturally embedded, and inherently relational. This paper proposes some expanded principles for trust in AI that can be incorporated into common development methods and frame trust as a dynamic, temporal relationship, which involves transparency and mutual respect. We draw on relational ethics and, in particular, African communitarian philosophies, to foreground the nuances of inclusive, participatory processes and long-term relationships with communities. Involving communities throughout the AI lifecycle can foster meaningful relationships with AI design and development teams that incrementally build trust and promote more equitable and context-sensitive AI systems. We illustrate how trust-enabling principles based on African relational ethics can be operationalised, using two use-cases for AI: healthcare and education.
Authors: Pit Neitemeier, Alessio Serra, Jiaze Li, Sascha Wirges, Lukas Balles, Jan Hendrik Metzen
Abstract: Hierarchical sequence models replace fixed tokenization with learned segmentations that compress long byte sequences for efficient autoregressive modeling. While recent end-to-end methods can learn meaningful boundaries from the language-modeling objective alone, it remains difficult to quantitatively assess and systematically steer where compute is spent. We introduce a router-agnostic metric of boundary quality, boundary enrichment B, which measures how strongly chunk starts concentrate on positions with high next-byte surprisal. Guided by this metric, we propose Sombrero, which steers boundary placement toward predictive difficulty via a confidence-alignment boundary loss and stabilizes boundary learning by applying confidence-weighted smoothing at the input level rather than on realized chunks. On 1B scale, across UTF-8 corpora covering English and German text as well as code and mathematical content, Sombrero improves the accuracy-efficiency trade-off and yields boundaries that more consistently align compute with hard-to-predict positions.
Authors: Charles Westphal, Keivan Navaie, Fernando E. Rosas
Abstract: Fine-tuned LLMs can covertly encode prompt secrets into outputs via steganographic channels. Prior work demonstrated this threat but relied on trivially recoverable encodings. We formalize payload recoverability via classifier accuracy and show previous schemes achieve 100\% recoverability. In response, we introduce low-recoverability steganography, replacing arbitrary mappings with embedding-space-derived ones. For Llama-8B (LoRA) and Ministral-8B (LoRA) trained on TrojanStego prompts, exact secret recovery rises from 17$\rightarrow$30\% (+78\%) and 24$\rightarrow$43\% (+80\%) respectively, while on Llama-70B (LoRA) trained on Wiki prompts, it climbs from 9$\rightarrow$19\% (+123\%), all while reducing payload recoverability. We then discuss detection. We argue that detecting fine-tuning-based steganographic attacks requires approaches beyond traditional steganalysis. Standard approaches measure distributional shift, which is an expected side-effect of fine-tuning. Instead, we propose a mechanistic interpretability approach: linear probes trained on later-layer activations detect the secret with up to 33\% higher accuracy in fine-tuned models compared to base models, even for low-recoverability schemes. This suggests that malicious fine-tuning leaves actionable internal signatures amenable to interpretability-based defenses.
Authors: Arya Hadizadeh Moghaddam, Mohsen Nayebi Kerdabadi, Dongjie Wang, Mei Liu, Zijun Yao
Abstract: Large-scale Electronic Health Record (EHR) databases have become indispensable in supporting clinical decision-making through data-driven treatment recommendations. However, existing medication recommender methods often struggle with a user (i.e., patient) cold-start problem, where recommendations for new patients are usually unreliable due to the lack of sufficient prescription history for patient profiling. While prior studies have utilized medical knowledge graphs to connect medication concepts through pharmacological or chemical relationships, these methods primarily focus on mitigating the item cold-start issue and fall short in providing personalized recommendations that adapt to individual patient characteristics. Meta-learning has shown promise in handling new users with sparse interactions in recommender systems. However, its application to EHRs remains underexplored due to the unique sequential structure of EHR data. To tackle these challenges, we propose MetaDrug, a multi-level, uncertainty-aware meta-learning framework designed to address the patient cold-start problem in medication recommendation. MetaDrug proposes a novel two-level meta-adaptation mechanism, including self-adaptation, which adapts the model to new patients using their own medical events as support sets to capture temporal dependencies; and peer-adaptation, which adapts the model using similar visits from peer patients to enrich new patient representations. Meanwhile, to further improve meta-adaptation outcomes, we introduce an uncertainty quantification module that ranks the support visits and filters out the unrelated information for adaptation consistency. We evaluate our approach on the MIMIC-III and Acute Kidney Injury (AKI) datasets. Experimental results on both datasets demonstrate that MetaDrug consistently outperforms state-of-the-art medication recommendation methods on cold-start patients.
Authors: Mathieu Petitbois, R\'emy Portelas, Sylvain Lamprier
Abstract: We study offline reinforcement learning of style-conditioned policies using explicit style supervision via subtrajectory labeling functions. In this setting, aligning style with high task performance is particularly challenging due to distribution shift and inherent conflicts between style and reward. Existing methods, despite introducing numerous definitions of style, often fail to reconcile these objectives effectively. To address these challenges, we propose a unified definition of behavior style and instantiate it into a practical framework. Building on this, we introduce Style-Conditioned Implicit Q-Learning (SCIQL), which leverages offline goal-conditioned RL techniques, such as hindsight relabeling and value learning, and combine it with a new Gated Advantage Weighted Regression mechanism to efficiently optimize task performance while preserving style alignment. Experiments demonstrate that SCIQL achieves superior performance on both objectives compared to prior offline methods. Code, datasets and visuals are available in: https://sciql-iclr-2026.github.io/.
Authors: Matthew Becker, Yifei Chen, Nicholas Cochran, Pouyan Ghasemi, Abhishek Gulati, Mark Harman, Zachary Haluza, Mehrdad Honarkhah, Herve Robert, Jiacheng Liu, Weini Liu, Sreeja Thummala, Xiaoning Yang, Rui Xin, Sophie Zeng
Abstract: We report on Just-in-Time catching test generation at Meta, designed to prevent bugs in large scale backend systems of hundreds of millions of line of code. Unlike traditional hardening tests, which pass at generation time, catching tests are meant to fail, surfacing bugs before code lands. The primary challenge is to reduce development drag from false positive test failures. Analyzing 22,126 generated tests, we show code-change-aware methods improve candidate catch generation 4x over hardening tests and 20x over coincidentally failing tests. To address false positives, we use rule-based and LLM-based assessors. These assessors reduce human review load by 70%. Inferential statistical analysis showed that human-accepted code changes are assessed to have significantly more false positives, while human-rejected changes have significantly more true positives. We reported 41 candidate catches to engineers; 8 were confirmed to be true positives, 4 of which would have led to serious failures had they remained uncaught. Overall, our results show that Just-in-Time catching is scalable, industrially applicable, and that it prevents serious failures from reaching production.
Authors: Thor Vestergaard Christiansen, Karran Pandey, Alba Reinders, Karan Singh, Morten Rieger Hannemose, J. Andreas B{\ae}rentzen
Abstract: We introduce Text Encoded Extrusion (TEE), a text-based representation that expresses mesh construction as sequences of face extrusions rather than polygon lists, and a method for generating 3D meshes from TEE using a large language model (LLM). By learning extrusion sequences that assemble a mesh, similar to the way artists create meshes, our approach naturally supports arbitrary output face counts and produces manifold meshes by design, in contrast to recent transformer-based models. The learnt extrusion sequences can also be applied to existing meshes - enabling editing in addition to generation. To train our model, we decompose a library of quadrilateral meshes with non-self-intersecting face loops into constituent loops, which can be viewed as their building blocks, and finetune an LLM on the steps for reassembling the meshes by performing a sequence of extrusions. We demonstrate that our representation enables reconstruction, novel shape synthesis, and the addition of new features to existing meshes.
Authors: Chuanzhe Guo, Jingjing Wu, Sijun He, Yang Chen, Zhaoqi Kuang, Shilong Fan, Bingjin Chen, Siqi Bao, Jing Liu, Hua Wu, Qingfu Zhu, Wanxiang Che, Haifeng Wang
Abstract: The evolution of Large Language Model (LLM) agents for software engineering (SWE) is constrained by the scarcity of verifiable datasets, a bottleneck stemming from the complexity of constructing executable environments across diverse languages. To address this, we introduce MEnvAgent, a Multi-language framework for automated Environment construction that facilitates scalable generation of verifiable task instances. MEnvAgent employs a multi-agent Planning-Execution-Verification architecture to autonomously resolve construction failures and integrates a novel Environment Reuse Mechanism that reduces computational overhead by incrementally patching historical environments. Evaluations on MEnvBench, a new benchmark comprising 1,000 tasks across 10 languages, demonstrate that MEnvAgent outperforms baselines, improving Fail-to-Pass (F2P) rates by 8.6% while reducing time costs by 43%. Additionally, we demonstrate the utility of MEnvAgent by constructing MEnvData-SWE, the largest open-source polyglot dataset of realistic verifiable Docker environments to date, alongside solution trajectories that enable consistent performance gains on SWE tasks across a wide range of models. Our code, benchmark, and dataset are available at https://github.com/ernie-research/MEnvAgent.
Authors: Chanwook Park, Brian Kim, Jiachen Guo, Wing Kam Liu
Abstract: Neural networks and machine learning models for uncertainty quantification suffer from limited scalability and poor reliability compared to their deterministic counterparts. In industry-scale active learning settings, where generating a single high-fidelity simulation may require days or weeks of computation and produce data volumes on the order of gigabytes, they quickly become impractical. This paper proposes a scalable and reliable Bayesian surrogate model, termed the Bayesian Interpolating Neural Network (B-INN). The B-INN combines high-order interpolation theory with tensor decomposition and alternating direction algorithm to enable effective dimensionality reduction without compromising predictive accuracy. We theoretically show that the function space of a B-INN is a subset of that of Gaussian processes, while its Bayesian inference exhibits linear complexity, $\mathcal{O}(N)$, with respect to the number of training samples. Numerical experiments demonstrate that B-INNs can be from 20 times to 10,000 times faster with a robust uncertainty estimation compared to Bayesian neural networks and Gaussian processes. These capabilities make B-INN a practical foundation for uncertainty-driven active learning in large-scale industrial simulations, where computational efficiency and robust uncertainty calibration are paramount.
Authors: Tanay Raghunandan Srinivasa, Vivek Deulkar, Jia Bhargava, Mohammad Hajiesmaili, Prashant Shenoy
Abstract: Battery energy storage systems are increasingly deployed as fast-responding resources for grid balancing services such as frequency regulation and for mitigating renewable generation uncertainty. However, repeated charging and discharging induces cycling degradation and reduces battery lifetime. This paper studies the real-time scheduling of a heterogeneous battery fleet that collectively tracks a stochastic balancing signal subject to per-battery ramp-rate and capacity constraints, while minimizing long-term cycling degradation. Cycling degradation is fundamentally path-dependent: it is determined by charge-discharge cycles formed by the state-of-charge (SoC) trajectory and is commonly quantified via rainflow cycle counting. This non-Markovian structure makes it difficult to express degradation as an additive per-time-step cost, complicating classical dynamic programming approaches. We address this challenge by formulating the fleet scheduling problem as a Markov decision process (MDP) with constrained action space and designing a dense proxy reward that provides informative feedback at each time step while remaining aligned with long-term cycle-depth reduction. To scale learning to large state-action spaces induced by fine-grained SoC discretization and asymmetric per-battery constraints, we develop a function-approximation reinforcement learning method using an Extreme Learning Machine (ELM) as a random nonlinear feature map combined with linear temporal-difference learning. We evaluate the proposed approach on a toy Markovian signal model and on a Markovian model trained from real-world regulation signal traces obtained from the University of Delaware, and demonstrate consistent reductions in cycle-depth occurrence and degradation metrics compared to baseline scheduling policies.
Authors: Alexander Loth, Martin Kappes, Marc-Oliver Pahl
Abstract: As foundation models (FMs) approach human-level fluency, distinguishing synthetic from organic content has become a key challenge for Trustworthy Web Intelligence. This paper presents JudgeGPT and RogueGPT, a dual-axis framework that decouples "authenticity" from "attribution" to investigate the mechanisms of human susceptibility. Analyzing 918 evaluations across five FMs (including GPT-4 and Llama-2), we employ Structural Causal Models (SCMs) as a principal framework for formulating testable causal hypotheses about detection accuracy. Contrary to partisan narratives, we find that political orientation shows a negligible association with detection performance ($r=-0.10$). Instead, "fake news familiarity" emerges as a candidate mediator ($r=0.35$), suggesting that exposure may function as adversarial training for human discriminators. We identify a "fluency trap" where GPT-4 outputs (HumanMachineScore: 0.20) bypass Source Monitoring mechanisms, rendering them indistinguishable from human text. These findings suggest that "pre-bunking" interventions should target cognitive source monitoring rather than demographic segmentation to ensure trustworthy information ecosystems.
Authors: Li Zhou, Hao Jiang, Junjie Li, Tianrui Wang, Haizhou Li
Abstract: Achieving precise and controllable emotional expression is crucial for producing natural and context-appropriate speech in text-to-speech (TTS) synthesis. However, many emotion-aware TTS systems, including large language model (LLM)-based designs, rely on scaling fixed emotion embeddings or external guidance, limiting their ability to model emotion-specific latent characteristics. To address this gap, we present EmoShift, a lightweight activation-steering framework incorporating a EmoSteer layer, which learns a steering vector for each target emotion in the output embedding space to capture its latent offset and maintain stable, appropriate expression across utterances and categories. With only 10M trainable parameters,less than 1/30 of full fine-tuning, EmoShift outperforms zero-shot and fully fine-tuned baselines in objective and subjective evaluations, enhancing emotional expressiveness while preserving naturalness and speaker similarity. Further analysis confirms the proposed EmoSteer layer's effectiveness and reveals its potential for controllable emotional intensity in speech synthesis.
Authors: Tanay Raghunandan Srinivasa, Vivek Deulkar, Aviruch Bhatia, Vishal Garg
Abstract: We study the joint operation and sizing of cooling infrastructure for commercial HVAC systems using reinforcement learning, with the objective of minimizing life-cycle cost over a 30-year horizon. The cooling system consists of a fixed-capacity electric chiller and a thermal energy storage (TES) unit, jointly operated to meet stochastic hourly cooling demands under time-varying electricity prices. The life-cycle cost accounts for both capital expenditure and discounted operating cost, including electricity consumption and maintenance. A key challenge arises from the strong asymmetry in capital costs: increasing chiller capacity by one unit is far more expensive than an equivalent increase in TES capacity. As a result, identifying the right combination of chiller and TES sizes, while ensuring zero loss-of-cooling-load under optimal operation, is a non-trivial co-design problem. To address this, we formulate the chiller operation problem for a fixed infrastructure configuration as a finite-horizon Markov Decision Process (MDP), in which the control action is the chiller part-load ratio (PLR). The MDP is solved using a Deep Q Network (DQN) with a constrained action space. The learned DQN RL policy minimizes electricity cost over historical traces of cooling demand and electricity prices. For each candidate chiller-TES sizing configuration, the trained policy is evaluated. We then restrict attention to configurations that fully satisfy the cooling demand and perform a life-cycle cost minimization over this feasible set to identify the cost-optimal infrastructure design. Using this approach, we determine the optimal chiller and thermal energy storage capacities to be 700 and 1500, respectively.
Authors: Yangyan Li
Abstract: Autoregressive sequence modeling stands as the cornerstone of modern Generative AI, powering results across diverse modalities ranging from text generation to image generation. However, a fundamental limitation of this paradigm is the rigid structural coupling of model capacity to computational cost: expanding a model's parametric memory -- its repository of factual knowledge or visual patterns -- traditionally requires deepening or widening the network, which incurs a proportional rise in active FLOPs. In this work, we introduce $\textbf{MoVE (Mixture of Value Embeddings)}$, a mechanism that breaks this coupling and establishes a new axis for scaling capacity. MoVE decouples memory from compute by introducing a global bank of learnable value embeddings shared across all attention layers. For every step in the sequence, the model employs a differentiable soft gating mechanism to dynamically mix retrieved concepts from this bank into the standard value projection. This architecture allows parametric memory to be scaled independently of network depth by simply increasing the number of embedding slots. We validate MoVE through strictly controlled experiments on two representative applications of autoregressive modeling: Text Generation and Image Generation. In both domains, MoVE yields consistent performance improvements over standard and layer-wise memory baselines, enabling the construction of "memory-dense" models that achieve lower perplexity and higher fidelity than their dense counterparts at comparable compute budgets.
Authors: Jio Oh, Paul Vicinanza, Thomas Butler, Steven Euijong Whang, Dezhi Hong, Amani Namboori
Abstract: More than 80% of the 1.6 billion English speakers do not use Standard American English (SAE) and experience higher failure rates and stereotyped responses when interacting with LLMs as a result. Yet multi-dialectal performance remains underexplored. We introduce $\textbf{MDial}$, the first large-scale framework for generating multi-dialectal conversational data encompassing the three pillars of written dialect -- lexical (vocabulary), orthographic (spelling), and morphosyntactic (grammar) features -- for nine English dialects. Partnering with native linguists, we design an annotated and scalable rule-based LLM transformation to ensure precision. Our approach challenges the assumption that models should mirror users' morphosyntactic features, showing that up to 90% of the grammatical features of a dialect should not be reproduced by models. Independent evaluations confirm data quality, with annotators preferring MDial outputs over prior methods in 98% of pairwise comparisons for dialect naturalness. Using this pipeline, we construct the dialect-parallel $\textbf{MDialBench}$mark with 50k+ dialogs, resulting in 97k+ QA pairs, and evaluate 17 LLMs on dialect identification and response generation tasks. Even frontier models achieve under 70% accuracy, fail to reach 50% for Canadian English, and systematically misclassify non-SAE dialects as American or British. As dialect identification underpins natural language understanding, these errors risk cascading failures into downstream tasks.
Authors: Yuxuan Lou, Ziming Wu, Yaochen Wang, Yong Liu, Yingxuan Ren, Fuming Lai, Shaobing Lian, Jie Tang, Yang You
Abstract: Current speech language models generate responses directly without explicit reasoning, leading to errors that cannot be corrected once audio is produced. We introduce \textbf{``Silent Thought, Spoken Answer''} -- a paradigm where speech LLMs generate internal text reasoning alongside spoken responses, with thinking traces informing speech quality. To realize this, we present \method{}, the first diffusion-based speech-text language model supporting both understanding and generation, unifying discrete text and tokenized speech under a single masked diffusion framework. Unlike autoregressive approaches, \method{} jointly generates reasoning traces and speech tokens through iterative denoising, with modality-specific masking schedules. We also construct \dataset{}, the first speech QA dataset with paired text reasoning traces, containing 26K samples totaling 319 hours. Experiments show \method{} achieves state-of-the-art speech-to-speech QA accuracy, outperforming the best baseline by up to 9 points, while attaining the best TTS quality among generative models (6.2\% WER) and preserving language understanding (66.2\% MMLU). Ablations confirm that both the diffusion architecture and thinking traces contribute to these gains.
Authors: Hun Chang, Byunghee Cha, Jong Chul Ye
Abstract: Recent studies have explored using pretrained Vision Foundation Models (VFMs) such as DINO for generative autoencoders, showing strong generative performance. Unfortunately, existing approaches often suffer from limited reconstruction fidelity due to the loss of high-frequency details. In this work, we present the DINO Spherical Autoencoder (DINO-SAE), a framework that bridges semantic representation and pixel-level reconstruction. Our key insight is that semantic information in contrastive representations is primarily encoded in the direction of feature vectors, while forcing strict magnitude matching can hinder the encoder from preserving fine-grained details. To address this, we introduce Hierarchical Convolutional Patch Embedding module that enhances local structure and texture preservation, and Cosine Similarity Alignment objective that enforces semantic consistency while allowing flexible feature magnitudes for detail retention. Furthermore, leveraging the observation that SSL-based foundation model representations intrinsically lie on a hypersphere, we employ Riemannian Flow Matching to train a Diffusion Transformer (DiT) directly on this spherical latent manifold. Experiments on ImageNet-1K demonstrate that our approach achieves state-of-the-art reconstruction quality, reaching 0.37 rFID and 26.2 dB PSNR, while maintaining strong semantic alignment to the pretrained VFM. Notably, our Riemannian Flow Matching-based DiT exhibits efficient convergence, achieving a gFID of 3.47 at 80 epochs.
Authors: Farnaz Soltaniani, Shoaib Razzaq, Mohammad Ghafari
Abstract: Early detection of security bug reports (SBRs) is critical for timely vulnerability mitigation. We present an evaluation of prompt-based engineering and fine-tuning approaches for predicting SBRs using Large Language Models (LLMs). Our findings reveal a distinct trade-off between the two approaches. Prompted proprietary models demonstrate the highest sensitivity to SBRs, achieving a G-measure of 77% and a recall of 74% on average across all the datasets, albeit at the cost of a higher false-positive rate, resulting in an average precision of only 22%. Fine-tuned models, by contrast, exhibit the opposite behavior, attaining a lower overall G-measure of 51% but substantially higher precision of 75% at the cost of reduced recall of 36%. Though a one-time investment in building fine-tuned models is necessary, the inference on the largest dataset is up to 50 times faster than that of proprietary models. These findings suggest that further investigations to harness the power of LLMs for SBR prediction are necessary.
Authors: Weiqin Yang, Bohao Wang, Zhenxiang Xu, Jiawei Chen, Shengjia Zhang, Jingbang Chen, Canghong Jin, Can Wang
Abstract: Recent years have witnessed a rapid surge in research leveraging Large Language Models (LLMs) for recommendation. These methods typically employ supervised fine-tuning (SFT) to adapt LLMs to recommendation scenarios, and utilize beam search during inference to efficiently retrieve $B$ top-ranked recommended items. However, we identify a critical training-inference inconsistency: while SFT optimizes the overall probability of positive items, it does not guarantee that such items will be retrieved by beam search even if they possess high overall probabilities. Due to the greedy pruning mechanism, beam search can prematurely discard a positive item once its prefix probability is insufficient. To address this inconsistency, we propose BEAR (Beam-SEarch-Aware Regularization), a novel fine-tuning objective that explicitly accounts for beam search behavior during training. Rather than directly simulating beam search for each instance during training, which is computationally prohibitive, BEAR enforces a relaxed necessary condition: each token in a positive item must rank within the top-$B$ candidate tokens at each decoding step. This objective effectively mitigates the risk of incorrect pruning while incurring negligible computational overhead compared to standard SFT. Extensive experiments across four real-world datasets demonstrate that BEAR significantly outperforms strong baselines. Code will be released upon acceptance.
Authors: Xidong Li, Mingyu Guo, Chenchao Xu, Bailin Li, Wenjing Zhu, Yangang Zou, Rui Chen, Zehuan Wang
Abstract: Trajectory planning is a core task in autonomous driving, requiring the prediction of safe and comfortable paths across diverse scenarios. Integrating Multi-modal Large Language Models (MLLMs) with Reinforcement Learning (RL) has shown promise in addressing "long-tail" scenarios. However, existing methods are constrained to single-turn reasoning, limiting their ability to handle complex tasks requiring iterative refinement. To overcome this limitation, we present MTDrive, a multi-turn framework that enables MLLMs to iteratively refine trajectories based on environmental feedback. MTDrive introduces Multi-Turn Group Relative Policy Optimization (mtGRPO), which mitigates reward sparsity by computing relative advantages across turns. We further construct an interactive trajectory understanding dataset from closed-loop simulation to support multi-turn training. Experiments on the NAVSIM benchmark demonstrate superior performance compared to existing methods, validating the effectiveness of our multi-turn reasoning paradigm. Additionally, we implement system-level optimizations to reduce data transfer overhead caused by high-resolution images and multi-turn sequences, achieving 2.5x training throughput. Our data, models, and code will be made available soon.
Authors: Evgeny Grigorenko, David Stanojevi\'c, David Ili\'c, Egor Bogomolov, Kostadin Cvejoski
Abstract: Modern Integrated Development Environments (IDEs) increasingly leverage Large Language Models (LLMs) to provide advanced features like code autocomplete. While powerful, training these models on user-written code introduces significant privacy risks, making the models themselves a new type of data vulnerability. Malicious actors can exploit this by launching attacks to reconstruct sensitive training data or infer whether a specific code snippet was used for training. This paper investigates the use of Differential Privacy (DP) as a robust defense mechanism for training an LLM for Kotlin code completion. We fine-tune a \texttt{Mellum} model using DP and conduct a comprehensive evaluation of its privacy and utility. Our results demonstrate that DP provides a strong defense against Membership Inference Attacks (MIAs), reducing the attack's success rate close to a random guess (AUC from 0.901 to 0.606). Furthermore, we show that this privacy guarantee comes at a minimal cost to model performance, with the DP-trained model achieving utility scores comparable to its non-private counterpart, even when trained on 100x less data. Our findings suggest that DP is a practical and effective solution for building private and trustworthy AI-powered IDE features.
Authors: Huan Song, Shuyu Tian, Junyi Hao, Cheng Yuan, Zhenyu Jia, Jiawei Shao, Xuelong Li
Abstract: As intelligent sensing expands into high-privacy environments such as restrooms and changing rooms, the field faces a critical privacy-security paradox. Traditional RGB surveillance raises significant concerns regarding visual recording and storage, while existing privacy-preserving methods-ranging from physical desensitization to traditional cryptographic or obfuscation techniques-often compromise semantic understanding capabilities or fail to guarantee mathematical irreversibility against reconstruction attacks. To address these challenges, this study presents a novel privacy-preserving perception technology based on the AI Flow theoretical framework and an edge-cloud collaborative architecture. The proposed methodology integrates source desensitization with irreversible feature mapping. Leveraging Information Bottleneck theory, the edge device performs millisecond-level processing to transform raw imagery into abstract feature vectors via non-linear mapping and stochastic noise injection. This process constructs a unidirectional information flow that strips identity-sensitive attributes, rendering the reconstruction of original images impossible. Subsequently, the cloud platform utilizes multimodal family models to perform joint inference solely on these abstract vectors to detect abnormal behaviors. This approach fundamentally severs the path to privacy leakage at the architectural level, achieving a breakthrough from video surveillance to de-identified behavior perception and offering a robust solution for risk management in high-sensitivity public spaces.
Authors: Farnaz Soltaniani, Mohammad Ghafari
Abstract: Machine learning models are increasingly used for software security tasks. These models are commonly trained and evaluated on large Internet-derived datasets, which often contain duplicated or highly similar samples. When such samples are split across training and test sets, data leakage may occur, allowing models to memorize patterns instead of learning to generalize. We investigate duplication in a widely used benchmark dataset of hard coded secrets and show how data leakage can substantially inflate the reported performance of AI-based secret detectors, resulting in a misleading picture of their real-world effectiveness.
Authors: Petar Veli\v{c}kovi\'c, Federico Barbero, Christos Perivolaropoulos, Simon Osindero, Razvan Pascanu
Abstract: Perplexity -- a function measuring a model's overall level of "surprise" when encountering a particular output -- has gained significant traction in recent years, both as a loss function and as a simple-to-compute metric of model quality. Prior studies have pointed out several limitations of perplexity, often from an empirical manner. Here we leverage recent results on Transformer continuity to show in a rigorous manner how perplexity may be an unsuitable metric for model selection. Specifically, we prove that, if there is any sequence that a compact decoder-only Transformer model predicts accurately and confidently -- a necessary pre-requisite for strong generalisation -- it must imply existence of another sequence with very low perplexity, but not predicted correctly by that same model. Further, by analytically studying iso-perplexity plots, we find that perplexity will not always select for the more accurate model -- rather, any increase in model confidence must be accompanied by a commensurate rise in accuracy for the new model to be selected.
Authors: Yuezhou Hu, Harman Singh, Monishwaran Maheswaran, Haocheng Xi, Coleman Hooper, Jintao Zhang, Aditya Tomar, Michael W. Mahoney, Sewon Min, Mehrdad Farajtabar, Kurt Keutzer, Amir Gholami, Chenfeng Xu
Abstract: Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to purely autoregressive language models because they can decode multiple tokens in parallel. However, state-of-the-art block-wise dLLMs rely on a "remasking" mechanism that decodes only the most confident tokens and discards the rest, effectively wasting computation. We demonstrate that recycling computation from the discarded tokens is beneficial, as these tokens retain contextual information useful for subsequent decoding iterations. In light of this, we propose Residual Context Diffusion (RCD), a module that converts these discarded token representations into contextual residuals and injects them back for the next denoising step. RCD uses a decoupled two-stage training pipeline to bypass the memory bottlenecks associated with backpropagation. We validate our method on both long CoT reasoning (SDAR) and short CoT instruction following (LLaDA) models. We demonstrate that a standard dLLM can be efficiently converted to the RCD paradigm with merely ~1 billion tokens. RCD consistently improves frontier dLLMs by 5-10 points in accuracy with minimal extra computation overhead across a wide range of benchmarks. Notably, on the most challenging AIME tasks, RCD nearly doubles baseline accuracy and attains up to 4-5x fewer denoising steps at equivalent accuracy levels.
Authors: Jeong Woon Lee, Kyoleen Kwak, Daeho Kim, Hyoseok Hwang
Abstract: Policies learned via continuous actor-critic methods often exhibit erratic, high-frequency oscillations, making them unsuitable for physical deployment. Current approaches attempt to enforce smoothness by directly regularizing the policy's output. We argue that this approach treats the symptom rather than the cause. In this work, we theoretically establish that policy non-smoothness is fundamentally governed by the differential geometry of the critic. By applying implicit differentiation to the actor-critic objective, we prove that the sensitivity of the optimal policy is bounded by the ratio of the Q-function's mixed-partial derivative (noise sensitivity) to its action-space curvature (signal distinctness). To empirically validate this theoretical insight, we introduce PAVE (Policy-Aware Value-field Equalization), a critic-centric regularization framework that treats the critic as a scalar field and stabilizes its induced action-gradient field. PAVE rectifies the learning signal by minimizing the Q-gradient volatility while preserving local curvature. Experimental results demonstrate that PAVE achieves smoothness and robustness comparable to policy-side smoothness regularization methods, while maintaining competitive task performance, without modifying the actor.
Authors: Wataru Uemura, Takeru Nagashima
Abstract: Factory automation has become increasingly important due to labor shortages, leading to the introduction of autonomous mobile robots for tasks such as material transportation. Markers are commonly used for robot self-localization and object identification. In the RoboCup Logistics League (RCLL), ArUco markers are employed both for robot localization and for identifying processing modules. Conventional recognition relies on OpenCV-based image processing, which detects black-and-white marker patterns. However, these methods often fail under noise, motion blur, defocus, or varying illumination conditions. Deep-learning-based recognition offers improved robustness under such conditions, but requires large amounts of annotated data. Annotation must typically be done manually, as the type and position of objects cannot be detected automatically, making dataset preparation a major bottleneck. In contrast, ArUco markers include built-in recognition modules that provide both ID and positional information, enabling automatic annotation. This paper proposes an automated annotation method for training deep-learning models on ArUco marker images. By leveraging marker detection results obtained from the ArUco module, the proposed approach eliminates the need for manual labeling. A YOLO-based model is trained using the automatically annotated dataset, and its performance is evaluated under various conditions. Experimental results demonstrate that the proposed method improves recognition performance compared with conventional image-processing techniques, particularly for images affected by blur or defocus. Automatic annotation also reduces human effort and ensures consistent labeling quality. Future work will investigate the relationship between confidence thresholds and recognition performance.
Authors: Yinsong Wang, Thomas Fletcher, Xinzhe Luo, Aine Travers Dineen, Rhodri Cusack, Chen Qin
Abstract: Reconstructing 3D fetal MR volumes from motion-corrupted stacks of 2D slices is a crucial and challenging task. Conventional slice-to-volume reconstruction (SVR) methods are time-consuming and require multiple orthogonal stacks for reconstruction. While learning-based SVR approaches have significantly reduced the time required at the inference stage, they heavily rely on ground truth information for training, which is inaccessible in practice. To address these challenges, we propose GaussianSVR, a self-supervised framework for slice-to-volume reconstruction. GaussianSVR represents the target volume using 3D Gaussian representations to achieve high-fidelity reconstruction. It leverages a simulated forward slice acquisition model to enable self-supervised training, alleviating the need for ground-truth volumes. Furthermore, to enhance both accuracy and efficiency, we introduce a multi-resolution training strategy that jointly optimizes Gaussian parameters and spatial transformations across different resolution levels. Experiments show that GaussianSVR outperforms the baseline methods on fetal MR volumetric reconstruction. Code will be available upon acceptance.
Authors: Yufei Gu, Zeke Xie
Abstract: While large language models (LLMs) have emerged as a significant advancement in artificial intelligence, the hardware and computational costs for training LLMs are also significantly burdensome. Among the state-of-the-art optimizers, AdamW relies on diagonal curvature estimates and ignores structural properties, while Muon applies global spectral normalization at the expense of losing curvature information. In this study, we restriked manifold optimization methods for training LLMs, which may address both optimizers' limitations, while conventional manifold optimization methods have been largely overlooked due to the poor performance in large-scale model optimization. By innovatively projecting the momentum onto the tangent space of model parameters and constraining it on a rotational Oblique manifold, we propose a novel, powerful, and efficient optimizer **Mano** that is the first to bridge the performance gap between manifold optimization and modern optimizers. Extensive experiments on the LLaMA and Qwen3 models demonstrate that Mano consistently and significantly outperforms AdamW and Muon even with less memory consumption and computational complexity, respectively, suggesting an expanded Pareto frontier in terms of space and time efficiency.
Authors: Afrozah Nadeem, Agrima, Mehwish Nasim, Usman Naseem
Abstract: Large Language Models (LLMs) increasingly shape global discourse, making fairness and ideological neutrality essential for responsible AI deployment. Despite growing attention to political bias in LLMs, prior work largely focuses on high-resource, Western languages or narrow multilingual settings, leaving cross-lingual consistency and safe post-hoc mitigation underexplored. To address this gap, we present a large-scale multilingual evaluation of political bias spanning 50 countries and 33 languages. We introduce a complementary post-hoc mitigation framework, Cross-Lingual Alignment Steering (CLAS), designed to augment existing steering methods by aligning ideological representations across languages and dynamically regulating intervention strength. This method aligns latent ideological representations induced by political prompts into a shared ideological subspace, ensuring cross lingual consistency, with the adaptive mechanism prevents over correction and preserves coherence. Experiments demonstrate substantial bias reduction along both economic and social axes with minimal degradation in response quality. The proposed framework establishes a scalable and interpretable paradigm for fairness-aware multilingual LLM governance, balancing ideological neutrality with linguistic and cultural diversity.
Authors: Xinchen Han, Qiuyang Fang, Hossam Afifi, Michel Marot
Abstract: Offline Reinforcement Learning (RL) relies on policy constraints to mitigate extrapolation error, where both the constraint form and constraint strength critically shape performance. However, most existing methods commit to a single constraint family: weighted behavior cloning, density regularization, or support constraints, without a unified principle that explains their connections or trade-offs. In this work, we propose Continuous Constraint Interpolation (CCI), a unified optimization framework in which these three constraint families arise as special cases along a common constraint spectrum. The CCI framework introduces a single interpolation parameter that enables smooth transitions and principled combinations across constraint types. Building on CCI, we develop Automatic Constraint Policy Optimization (ACPO), a practical primal--dual algorithm that adapts the interpolation parameter via a Lagrangian dual update. Moreover, we establish a maximum-entropy performance difference lemma and derive performance lower bounds for both the closed-form optimal policy and its parametric projection. Experiments on D4RL and NeoRL2 demonstrate robust gains across diverse domains, achieving state-of-the-art performance overall.
Authors: Blagoj Hristov, Zoran Hadzi-Velkov, Katerina Hadzi-Velkova Saneva, Gorjan Nadzinski, Vesna Ojleska Latkoska
Abstract: Reliable control of myoelectric prostheses is often hindered by high inter-subject variability and the clinical impracticality of high-density sensor arrays. This study proposes a deep learning framework for accurate gesture recognition using only two surface electromyography (sEMG) channels. The method employs a Convolutional Sparse Autoencoder (CSAE) to extract temporal feature representations directly from raw signals, eliminating the need for heuristic feature engineering. On a 6-class gesture set, our model achieved a multi-subject F1-score of 94.3% $\pm$ 0.3%. To address subject-specific differences, we present a few-shot transfer learning protocol that improved performance on unseen subjects from a baseline of 35.1% $\pm$ 3.1% to 92.3% $\pm$ 0.9% with minimal calibration data. Furthermore, the system supports functional extensibility through an incremental learning strategy, allowing for expansion to a 10-class set with a 90.0% $\pm$ 0.2% F1-score without full model retraining. By combining high precision with minimal computational and sensor overhead, this framework provides a scalable and efficient approach for the next generation of affordable and adaptive prosthetic systems.
Authors: Yizhi Liu
Abstract: Differentiable matching layers, often implemented via entropy-regularized Optimal Transport, serve as a critical approximate inference mechanism in structural prediction. However, recovering discrete permutations via annealing $\epsilon \to 0$ is notoriously unstable. We identify a fundamental mechanism for this failure: \textbf{Premature Mode Collapse}. By analyzing the non-normal dynamics of the Sinkhorn fixed-point map, we reveal a theoretical \textbf{thermodynamic speed limit}. Under standard exponential cooling, the shift in the target posterior ($O(1)$) outpaces the contraction rate of the inference operator, which degrades as $O(1/\epsilon)$. This mismatch inevitably forces the inference trajectory into spurious local basins. To address this, we propose \textbf{Efficient PH-ASC}, an adaptive scheduling algorithm that monitors the stability of the inference process. By enforcing a linear stability law, we decouple expensive spectral diagnostics from the training loop, reducing overhead from $O(N^3)$ to amortized $O(1)$. Our implementation and interactive demo are available at https://github.com/xxx0438/torch-sinkhorn-asc and https://huggingface.co/spaces/leon0923/torch-sinkhorn-asc-demo. bounded away from zero in generic training dynamics unless the feature extractor converges unrealistically fast.
URLs: https://github.com/xxx0438/torch-sinkhorn-asc, https://huggingface.co/spaces/leon0923/torch-sinkhorn-asc-demo.
Authors: Seyedeh Ava Razi Razavi, James Sargant, Sheridan Houghten, Renata Dividino
Abstract: Generating realistic graph-structured data is challenging due to discrete structures, variable sizes, and class-specific connectivity patterns that resist conventional generative modelling. While recent graph generation methods employ generative adversarial network (GAN) frameworks to handle permutation invariance and irregular topologies, they typically rely on random edge sampling with fixed probabilities, limiting their capacity to capture complex structural dependencies between nodes. We propose a density-aware conditional graph generation framework using Wasserstein GANs (WGAN) that replaces random sampling with a learnable distance-based edge predictor. Our approach embeds nodes into a latent space where proximity correlates with edge likelihood, enabling the generator to learn meaningful connectivity patterns. A differentiable edge predictor determines pairwise relationships directly from node embeddings, while a density-aware selection mechanism adaptively controls edge density to match class-specific sparsity distributions observed in real graphs. We train the model using a WGAN with gradient penalty, employing a GCN-based critic to ensure generated graphs exhibit realistic topology and align with target class distributions. Experiments on benchmark datasets demonstrate that our method produces graphs with superior structural coherence and class-consistent connectivity compared to existing baselines. The learned edge predictor captures complex relational patterns beyond simple heuristics, generating graphs whose density and topology closely match real structural distributions. Our results show improved training stability and controllable synthesis, making the framework effective for realistic graph generation and data augmentation. Source code is publicly available at https://github.com/ava-12/Density_Aware_WGAN.git.
Authors: Antonio Vitale, Emanuela Guglielmi, Simone Scalabrino, Rocco Oliveto
Abstract: Large Language Models (LLMs) are increasingly relevant in Software Engineering research and practice, with Automated Bug Fixing (ABF) being one of their key applications. ABF involves transforming a buggy method into its fixed equivalent. A common preprocessing step in ABF involves removing comments from code prior to training. However, we hypothesize that comments may play a critical role in fixing certain types of bugs by providing valuable design and implementation insights. In this study, we investigate how the presence or absence of comments, both during training and at inference time, impacts the bug-fixing capabilities of LLMs. We conduct an empirical evaluation comparing two model families, each evaluated under all combinations of training and inference conditions (with and without comments), and thereby revisiting the common practice of removing comments during training. To address the limited availability of comments in state-of-the-art datasets, we use an LLM to automatically generate comments for methods lacking them. Our findings show that comments improve ABF accuracy by up to threefold when present in both phases, while training with comments does not degrade performance when instances lack them. Additionally, an interpretability analysis identifies that comments detailing method implementation are particularly effective in aiding LLMs to fix bugs accurately.
Authors: Hari Krishna Gadi, Daniel Matos, Hongyi Luo, Lu Liu, Yongliang Wang, Yanfeng Zhang, Liqiu Meng
Abstract: Visual geolocalization, the task of predicting where an image was taken, remains challenging due to global scale, visual ambiguity, and the inherently hierarchical structure of geography. Existing paradigms rely on either large-scale retrieval, which requires storing a large number of image embeddings, grid-based classifiers that ignore geographic continuity, or generative models that diffuse over space but struggle with fine detail. We introduce an entity-centric formulation of geolocation that replaces image-to-image retrieval with a compact hierarchy of geographic entities embedded in Hyperbolic space. Images are aligned directly to country, region, subregion, and city entities through Geo-Weighted Hyperbolic contrastive learning by directly incorporating haversine distance into the contrastive objective. This hierarchical design enables interpretable predictions and efficient inference with 240k entity embeddings instead of over 5 million image embeddings on the OSV5M benchmark, on which our method establishes a new state-of-the-art performance. Compared to the current methods in the literature, it reduces mean geodesic error by 19.5\%, while improving the fine-grained subregion accuracy by 43%. These results demonstrate that geometry-aware hierarchical embeddings provide a scalable and conceptually new alternative for global image geolocation.
Authors: Xiaoxuan Guo, Yuankun Xie, Haonan Cheng, Jiayi Zhou, Jian Liu, Hengyan Huang, Long Ye, Qin Zhang
Abstract: Speech deepfake detection (SDD) focuses on identifying whether a given speech signal is genuine or has been synthetically generated. Existing audio large language model (LLM)-based methods excel in content understanding; however, their predictions are often biased toward semantically correlated cues, which results in fine-grained acoustic artifacts being overlooked during the decisionmaking process. Consequently, fake speech with natural semantics can bypass detectors despite harboring subtle acoustic anomalies; this suggests that the challenge stems not from the absence of acoustic data, but from its inadequate accessibility when semantic-dominant reasoning prevails. To address this issue, we investigate SDD within the audio LLM paradigm and introduce SDD with Auditory Perception-enhanced Audio Large Language Model (SDD-APALLM), an acoustically enhanced framework designed to explicitly expose fine-grained time-frequency evidence as accessible acoustic cues. By combining raw audio with structured spectrograms, the proposed framework empowers audio LLMs to more effectively capture subtle acoustic inconsistencies without compromising their semantic understanding. Experimental results indicate consistent gains in detection accuracy and robustness, especially in cases where semantic cues are misleading. Further analysis reveals that these improvements stem from a coordinated utilization of semantic and acoustic information, as opposed to simple modality aggregation.
Authors: Joao Fonseca, Julia Stoyanovich
Abstract: Computing the importance of features in supervised classification tasks is critical for model interpretability. Shapley values are a widely used approach for explaining model predictions, but require direct access to the underlying model, an assumption frequently violated in real-world deployments. Further, even when model access is possible, their exact computation may be prohibitively expensive. We investigate whether meaningful Shapley value estimations can be obtained in a zero-shot setting, using only the input data distribution and no evaluations of the target model. To this end, we introduce ExplainerPFN, a tabular foundation model built on TabPFN that is pretrained on synthetic datasets generated from random structural causal models and supervised using exact or near-exact Shapley values. Once trained, ExplainerPFN predicts feature attributions for unseen tabular datasets without model access, gradients, or example explanations. Our contributions are fourfold: (1) we show that few-shot learning-based explanations can achieve high fidelity to SHAP values with as few as two reference observations; (2) we propose ExplainerPFN, the first zero-shot method for estimating Shapley values without access to the underlying model or reference explanations; (3) we provide an open-source implementation of ExplainerPFN, including the full training pipeline and synthetic data generator; and (4) through extensive experiments on real and synthetic datasets, we show that ExplainerPFN achieves performance competitive with few-shot surrogate explainers that rely on 2-10 SHAP examples.
Authors: Yanghao Su, Wenbo Zhou, Tianwei Zhang, Qiu Han, Weiming Zhang, Nenghai Yu, Jie Zhang
Abstract: Emergent Misalignment refers to a failure mode in which fine-tuning large language models (LLMs) on narrowly scoped data induces broadly misaligned behavior. Prior explanations mainly attribute this phenomenon to the generalization of erroneous or unsafe content. In this work, we show that this view is incomplete. Across multiple domains and model families, we find that fine-tuning models on data exhibiting specific character-level dispositions induces substantially stronger and more transferable misalignment than incorrect-advice fine-tuning, while largely preserving general capabilities. This indicates that emergent misalignment arises from stable shifts in model behavior rather than from capability degradation or corrupted knowledge. We further show that such behavioral dispositions can be conditionally activated by both training-time triggers and inference-time persona-aligned prompts, revealing shared structure across emergent misalignment, backdoor activation, and jailbreak susceptibility. Overall, our results identify character formation as a central and underexplored alignment risk, suggesting that robust alignment must address behavioral dispositions rather than isolated errors or prompt-level defenses.
Authors: Mohanna Hoveyda, Jelle Piepenbrock, Arjen P de Vries, Maarten de Rijke, Faegheh Hasibi
Abstract: Resolving complex information needs that come with multiple constraints should consider enforcing the logical operators encoded in the query (i.e., conjunction, disjunction, negation) on the candidate answer set. Current retrieval systems either ignore these constraints in neural embeddings or approximate them in a generative reasoning process that can be inconsistent and unreliable. Although well-suited to structured reasoning, existing neuro-symbolic approaches remain confined to formal logic or mathematics problems as they often assume unambiguous queries and access to complete evidence, conditions rarely met in information retrieval. To bridge this gap, we introduce OrLog, a neuro-symbolic retrieval framework that decouples predicate-level plausibility estimation from logical reasoning: a large language model (LLM) provides plausibility scores for atomic predicates in one decoding-free forward pass, from which a probabilistic reasoning engine derives the posterior probability of query satisfaction. We evaluate OrLog across multiple backbone LLMs, varying levels of access to external knowledge, and a range of logical constraints, and compare it against base retrievers and LLM-as-reasoner methods. Provided with entity descriptions, OrLog can significantly boost top-rank precision compared to LLM reasoning with larger gains on disjunctive queries. OrLog is also more efficient, cutting mean tokens by $\sim$90\% per query-entity pair. These results demonstrate that generation-free predicate plausibility estimation combined with probabilistic reasoning enables constraint-aware retrieval that outperforms monolithic reasoning while using far fewer tokens.
Authors: Zhixiang Zhang, Zesen Liu, Yuchong Xie, Quanfeng Huang, Dongdong She
Abstract: Semantic caching has emerged as a pivotal technique for scaling LLM applications, widely adopted by major providers including AWS and Microsoft. By utilizing semantic embedding vectors as cache keys, this mechanism effectively minimizes latency and redundant computation for semantically similar queries. In this work, we conceptualize semantic cache keys as a form of fuzzy hashes. We demonstrate that the locality required to maximize cache hit rates fundamentally conflicts with the cryptographic avalanche effect necessary for collision resistance. Our conceptual analysis formalizes this inherent trade-off between performance (locality) and security (collision resilience), revealing that semantic caching is naturally vulnerable to key collision attacks. While prior research has focused on side-channel and privacy risks, we present the first systematic study of integrity risks arising from cache collisions. We introduce CacheAttack, an automated framework for launching black-box collision attacks. We evaluate CacheAttack in security-critical tasks and agentic workflows. It achieves a hit rate of 86\% in LLM response hijacking and can induce malicious behaviors in LLM agent, while preserving strong transferability across different embedding models. A case study on a financial agent further illustrates the real-world impact of these vulnerabilities. Finally, we discuss mitigation strategies.
Authors: Haitham S. Al-Sinani, Chris J. Mitchell
Abstract: Wireless ethical hacking relies heavily on skilled practitioners manually interpreting reconnaissance results and executing complex, time-sensitive sequences of commands to identify vulnerable targets, capture authentication handshakes, and assess password resilience; a process that is inherently labour-intensive, difficult to scale, and prone to subjective judgement and human error. To help address these limitations, we propose WiFiPenTester, an experimental, governed, and reproducible system for GenAI-enabled wireless ethical hacking. The system integrates large language models into the reconnaissance and decision-support phases of wireless security assessment, enabling intelligent target ranking, attack feasibility estimation, and strategy recommendation, while preserving strict human-in-the-loop control and budget-aware execution. We describe the system architecture, threat model, governance mechanisms, and prompt-engineering methodology, and empirical experiments conducted across multiple wireless environments. The results demonstrate that GenAI assistance improves target selection accuracy and overall assessment efficiency, while maintaining auditability and ethical safeguards. This indicates that WiFiPenTester is a meaningful step toward practical, safe, and scalable GenAI-assisted wireless penetration testing, while reinforcing the necessity of bounded autonomy, human oversight, and rigorous governance mechanisms when deploying GenAI in ethical hacking.
Authors: Jiaming Ma, Siyuan Mu, Ruilin Tang, Haofeng Ma, Qihe Huang, Zhengyang Zhou, Pengkun Wang, Binwu Wang, Yang Wang
Abstract: The prevailing Direct Forecasting (DF) paradigm dominates Long-term Time Series Forecasting (LTSF) by forcing models to predict the entire future horizon in a single forward pass. While efficient, this rigid coupling of output and evaluation horizons necessitates computationally prohibitive re-training for every target horizon. In this work, we uncover a counter-intuitive optimization anomaly: models trained on short horizons-when coupled with our proposed Evolutionary Forecasting (EF) paradigm-significantly outperform those trained directly on long horizons. We attribute this success to the mitigation of a fundamental optimization pathology inherent in DF, where conflicting gradients from distant futures cripple the learning of local dynamics. We establish EF as a unified generative framework, proving that DF is merely a degenerate special case of EF. Extensive experiments demonstrate that a singular EF model surpasses task-specific DF ensembles across standard benchmarks and exhibits robust asymptotic stability in extreme extrapolation. This work propels a paradigm shift in LTSF: moving from passive Static Mapping to autonomous Evolutionary Reasoning.
Authors: Christiaan P. Opperman, Anna S. Bosman, Katherine M. Malan
Abstract: Despite huge successes on a wide range of tasks, neural networks are known to sometimes struggle to generalise to unseen data. Many approaches have been proposed over the years to promote the generalisation ability of neural networks, collectively known as regularisation techniques. These are used as common practice under the assumption that any regularisation added to the pipeline would result in a performance improvement. In this study, we investigate whether this assumption holds in practice. First, we provide a broad review of regularisation techniques, including modern theories such as double descent. We propose a taxonomy of methods under four broad categories, namely: (1) data-based strategies, (2) architecture strategies, (3) training strategies, and (4) loss function strategies. Notably, we highlight the contradictions and correspondences between the approaches in these broad classes. Further, we perform an empirical comparison of the various regularisation techniques on classification tasks for ten numerical and image datasets applied to the multi-layer perceptron and convolutional neural network architectures. Results show that the efficacy of regularisation is dataset-dependent. For example, the use of a regularisation term only improved performance on numeric datasets, whereas batch normalisation improved performance on image datasets only. Generalisation is crucial to machine learning; thus, understanding the effects of applying regularisation techniques, and considering the connections between them is essential to the appropriate use of these methods in practice.
Authors: Saeid Jamshidi, Kawser Wazed Nafi, Arghavan Moradi Dakhel, Foutse Khomh, Amin Nikanjam, Mohammad Adnan Hamdaqa
Abstract: Large Language Models (LLMs) are increasingly adopted in sensitive domains such as healthcare and financial institutions' data analytics; however, their execution pipelines remain vulnerable to manipulation and unverifiable behavior. Existing control mechanisms, such as the Model Context Protocol (MCP), define compliance policies for tool invocation but lack verifiable enforcement and transparent validation of model actions. To address this gap, we propose a novel Secure Tool Manifest and Digital Signing Framework, a structured and security-aware extension of Model Context Protocols. The framework enforces cryptographically signed manifests, integrates transparent verification logs, and isolates model-internal execution metadata from user-visible components to ensure verifiable execution integrity. Furthermore, the evaluation demonstrates that the framework scales nearly linearly (R-squared = 0.998), achieves near-perfect acceptance of valid executions while consistently rejecting invalid ones, and maintains balanced model utilization across execution pipelines.
Authors: Zheyuan Hu, Yifei Shi
Abstract: In the post-Dennard era, optimizing embedded systems requires navigating complex trade-offs between energy efficiency and latency. Traditional heuristic tuning is often inefficient in such high-dimensional, non-smooth landscapes. In this work, we propose a Bayesian Optimization framework using Gaussian Processes to automate the search for optimal scheduling configurations on heterogeneous multi-core architectures. We explicitly address the multi-objective nature of the problem by approximating the Pareto Frontier between energy and time. Furthermore, by incorporating Sensitivity Analysis (fANOVA) and comparing different covariance kernels (e.g., Mat\'ern vs. RBF), we provide physical interpretability to the black-box model, revealing the dominant hardware parameters driving system performance.
Authors: Saeid Jamshidi, Omar Abdul Wahab, Rolando Herrero, Foutse Khomh
Abstract: The integrity of time in distributed Internet of Things (IoT) devices is crucial for reliable operation in energy cyber-physical systems, such as smart grids and microgrids. However, IoT systems are vulnerable to clock drift, time-synchronization manipulation, and timestamp discontinuities, such as the Year 2038 (Y2K38) Unix overflow, all of which disrupt temporal ordering. Conventional anomaly-detection models, which assume reliable timestamps, fail to capture temporal inconsistencies. This paper introduces STGAT (Spatio-Temporal Graph Attention Network), a framework that models both temporal distortion and inter-device consistency in energy IoT systems. STGAT combines drift-aware temporal embeddings and temporal self-attention to capture corrupted time evolution at individual devices, and uses graph attention to model spatial propagation of timing errors. A curvature-regularized latent representation geometrically separates normal clock evolution from anomalies caused by drift, synchronization offsets, and overflow events. Experimental results on energy IoT telemetry with controlled timing perturbations show that STGAT achieves 95.7% accuracy, outperforming recurrent, transformer, and graph-based baselines with significant improvements (d > 1.8, p < 0.001). Additionally, STGAT reduces detection delay by 26%, achieving a 2.3-time-step delay while maintaining stable performance under overflow, drift, and physical inconsistencies.
Authors: Joel Romero-Hernandez, Oscar Camara
Abstract: Pain management in intensive care usually involves complex trade-offs between therapeutic goals and patient safety, since both inadequate and excessive treatment may induce serious sequelae. Reinforcement learning can help address this challenge by learning medication dosing policies from retrospective data. However, prior work on sedation and analgesia has optimized for objectives that do not value patient survival while relying on algorithms unsuitable for imperfect information settings. We investigated the risks of these design choices by implementing a deep reinforcement learning framework to suggest hourly medication doses under partial observability. Using data from 47,144 ICU stays in the MIMIC-IV database, we trained policies to prescribe opioids, propofol, benzodiazepines, and dexmedetomidine according to two goals: reduce pain or jointly reduce pain and mortality. We found that, although the two policies were associated with lower pain, actions from the first policy were positively correlated with mortality, while those proposed by the second policy were negatively correlated. This suggests that valuing long-term outcomes could be critical for safer treatment policies, even if a short-term goal remains the primary objective.
Authors: Powei Chang, Jinpeng Zhang, Bowen Chen, Chenyu Wang, Chenlu Guo, Yixing Zhang, Yukang Gao, JianXiang Xiang, Yue Gao, Chaoqun Sun, Yiyi Chen, Dongying Kong
Abstract: Information-based data selection for instruction tuning is compelling: maximizing the log-determinant of the Fisher information yields a monotone submodular objective, enabling greedy algorithms to achieve a $(1-1/e)$ approximation under a cardinality budget. In practice, however, we identify alleviating gradient conflicts, misalignment between per-sample gradients, is a key factor that slows down the decay of marginal log-determinant information gains, thereby preventing significant loss of information. We formalize this via an $\varepsilon$-decomposition that quantifies the deviation from ideal submodularity as a function of conflict statistics, yielding data-dependent approximation factors that tighten as conflicts diminish. Guided by this analysis, we propose SPICE, a conflict-aware selector that maximizes information while penalizing misalignment, and that supports early stopping and proxy models for efficiency. Empirically, SPICE selects subsets with higher log-determinant information than original criteria, and these informational gains translate into performance improvements: across 8 benchmarks with LLaMA2-7B and Qwen2-7B, SPICE uses only 10% of the data, yet matches or exceeds 6 methods including full-data tuning. This achieves performance improvements with substantially lower training cost.
Authors: Marthe Ballon, Brecht Verbeken, Vincent Ginis, Andres Algaba
Abstract: Large language models (LLMs) increasingly solve difficult problems by producing "reasoning traces" before emitting a final response. However, it remains unclear how accuracy and decision commitment evolve along a reasoning trajectory, and whether intermediate trace segments provide answer-relevant information beyond generic length or stylistic effects. Here, we propose a protocol to systematically probe the trajectories of reasoning traces in LLMs by 1) generating a model's reasoning trace, 2) truncating it at fixed token-percentiles, and 3) injecting each partial trace back into the model (or a different model) to measure the induced distribution over answer choices via next-token probabilities. We apply this protocol to the open-source Qwen3-4B/-8B/-14B and gpt-oss-20b/-120b models across the multiple-choice GPQA Diamond and MMLU-Pro benchmarks. We find that accuracy and decision commitment consistently increase as the percentage of provided reasoning tokens grows. These gains are primarily driven by relevant content in the model generation rather than context length or generic "reasoning style" effects. Stronger models often backtrack successfully from incorrect partial traces, but immediate answers often remain anchored in the weaker model's incorrect response. More broadly, we show that trajectory probing provides diagnostics for efficient and safer deployment of reasoning models as the measurements can inform practical trace-handling and monitoring policies that improve reliability without assuming intermediate tokens are inherently faithful explanations.
Authors: Luca Della Libera, Cem Subakan, Mirco Ravanelli
Abstract: Neural audio codecs are at the core of modern conversational speech technologies, converting continuous speech into sequences of discrete tokens that can be processed by LLMs. However, existing codecs typically operate at fixed frame rates, allocating tokens uniformly in time and producing unnecessarily long sequences. In this work, we introduce DyCAST, a Dynamic Character-Aligned Speech Tokenizer that enables variable-frame-rate tokenization through soft character-level alignment and explicit duration modeling. DyCAST learns to associate tokens with character-level linguistic units during training and supports alignment-free inference with direct control over token durations at decoding time. To improve speech resynthesis quality at low frame rates, we further introduce a retrieval-augmented decoding mechanism that enhances reconstruction fidelity without increasing bitrate. Experiments show that DyCAST achieves competitive speech resynthesis quality and downstream performance while using significantly fewer tokens than fixed-frame-rate codecs.
Authors: Muhammad Fetrat Qharabagh, Artur Back de Luca, George Giapitzakis, Kimon Fountoulakis
Abstract: Understanding what graph neural networks can learn, especially their ability to learn to execute algorithms, remains a central theoretical challenge. In this work, we prove exact learnability results for graph algorithms under bounded-degree and finite-precision constraints. Our approach follows a two-step process. First, we train an ensemble of multi-layer perceptrons (MLPs) to execute the local instructions of a single node. Second, during inference, we use the trained MLP ensemble as the update function within a graph neural network (GNN). Leveraging Neural Tangent Kernel (NTK) theory, we show that local instructions can be learned from a small training set, enabling the complete graph algorithm to be executed during inference without error and with high probability. To illustrate the learning power of our setting, we establish a rigorous learnability result for the LOCAL model of distributed computation. We further demonstrate positive learnability results for widely studied algorithms such as message flooding, breadth-first and depth-first search, and Bellman-Ford.
Authors: Joshua Southern, Changpeng Lu, Santrupti Nerli, Samuel D. Stanton, Andrew M. Watkins, Franziska Seeger, Fr\'ed\'eric A. Dreyer
Abstract: Multispecific antibodies offer transformative therapeutic potential by engaging multiple epitopes simultaneously, yet their efficacy is an emergent property governed by complex molecular architectures. Rational design is often bottlenecked by the inability to predict how subtle changes in domain topology influence functional outcomes, a challenge exacerbated by the scarcity of comprehensive experimental data. Here, we introduce a computational framework to address part of this gap. First, we present a generative method for creating large-scale, realistic synthetic functional landscapes that capture non-linear interactions where biological activity depends on domain connectivity. Second, we propose a graph neural network architecture that explicitly encodes these topological constraints, distinguishing between format configurations that appear identical to sequence-only models. We demonstrate that this model, trained on synthetic landscapes, recapitulates complex functional properties and, via transfer learning, has the potential to achieve high predictive accuracy on limited biological datasets. We showcase the model's utility by optimizing trade-offs between efficacy and toxicity in trispecific T-cell engagers and retrieving optimal common light chains. This work provides a robust benchmarking environment for disentangling the combinatorial complexity of multispecifics, accelerating the design of next-generation therapeutics.
Authors: Shuai Shao, Yixiang Liu, Bingwei Lu, Weinan Zhang
Abstract: In recent years, LLM-based multi-agent systems (MAS) have advanced rapidly, using a router to decompose tasks and delegate subtasks to specialized agents. A natural way to expand capability is to scale up the agent pool by continually integrating new functional agents or tool interfaces, but naive expansion can trigger performance collapse when the router cold-starts on newly added, heterogeneous, and unreliable agents. We propose MonoScale, an expansion-aware update framework that proactively generates a small set of agent-conditioned familiarization tasks, harvests evidence from both successful and failed interactions, and distills it into auditable natural-language memory to guide future routing. We formalize sequential augmentation as a contextual bandit and perform trust-region memory updates, yielding a monotonic non-decreasing performance guarantee across onboarding rounds. Experiments on GAIA and Humanity's Last Exam show stable gains as the agent pool grows, outperforming naive scale-up and strong-router fixed-pool baselines.
Authors: Anglin Liu, Ruichao Chen, Yi Lu, Hongxia Xu, Jintai Chen
Abstract: Despite recent Multimodal Large Language Models (MLLMs)' linguistic prowess in medical diagnosis, we find even state-of-the-art MLLMs suffer from a critical perceptual deficit: geometric blindness. This failure to ground outputs in objective geometric constraints leads to plausible yet factually incorrect hallucinations, rooted in training paradigms that prioritize linguistic fluency over geometric fidelity. This paper introduces Med-Scout, a novel framework that "cures" this blindness via Reinforcement Learning (RL) that leverages the intrinsic geometric logic latent within unlabeled medical images. Instead of relying on costly expert annotations, Med-Scout derives verifiable supervision signals through three strategic proxy tasks: Hierarchical Scale Localization, Topological Jigsaw Reconstruction, and Anomaly Consistency Detection. To rigorously quantify this deficit, we present Med-Scout-Bench, a new benchmark specifically designed to evaluate geometric perception. Extensive evaluations show that Med-Scout significantly mitigates geometric blindness, outperforming leading proprietary and open-source MLLMs by over 40% on our benchmark. Furthermore, this enhanced geometric perception generalizes to broader medical understanding, achieving superior results on radiological and comprehensive medical VQA tasks.
Authors: Rajib Mostakim, Reza T. Batley, Sourav Saha
Abstract: Deep reinforcement learning (RL) is increasingly deployed in resource-constrained environments, yet the go-to function approximators - multilayer perceptrons (MLPs) - are often parameter-inefficient due to an imperfect inductive bias for the smooth structure of many value functions. This mismatch can also hinder sample efficiency and slow policy learning in this capacity-limited regime. Although model compression techniques exist, they operate post-hoc and do not improve learning efficiency. Recent spline-based separable architectures - such as Kolmogorov-Arnold Networks (KANs) - have been shown to offer parameter efficiency but are widely reported to exhibit significant computational overhead, especially at scale. In seeking to address these limitations, this work introduces SPAN (SPline-based Adaptive Networks), a novel function approximation approach to RL. SPAN adapts the low rank KHRONOS framework by integrating a learnable preprocessing layer with a separable tensor product B-spline basis. SPAN is evaluated across discrete (PPO) and high-dimensional continuous (SAC) control tasks, as well as offline settings (Minari/D4RL). Empirical results demonstrate that SPAN achieves a 30-50% improvement in sample efficiency and 1.3-9 times higher success rates across benchmarks compared to MLP baselines. Furthermore, SPAN demonstrates superior anytime performance and robustness to hyperparameter variations, suggesting it as a viable, high performance alternative for learning intrinsically efficient policies in resource-limited settings.
Authors: Tao Yu, Haopeng Jin, Hao Wang, Shenghua Chai, Yujia Yang, Junhao Gong, Jiaming Guo, Minghui Zhang, Xinlong Chen, Zhenghao Zhang, Yuxuan Zhou, Yanpei Gong, YuanCheng Liu, Yiming Ding, Kangwei Zeng, Pengfei Yang, Zhongtian Luo, Yufei Xiong, Shanbin Zhang, Shaoxiong Cheng, Huang Ruilin, Li Shuo, Yuxi Niu, Xinyuan Zhang, Yueya Xu, Jie Mao, Ruixuan Ji, Yaru Zhao, Mingchen Zhang, Jiabing Yang, Jiaqi Liu, YiFan Zhang, Hongzhu Yi, Xinming Wang, Cheng Zhong, Xiao Ma, Zhang Zhang, Yan Huang, Liang Wang
Abstract: In recent years, large language models (LLMs) have made rapid progress in information retrieval, yet existing research has mainly focused on text or static multimodal settings. Open-domain video shot retrieval, which involves richer temporal structure and more complex semantics, still lacks systematic benchmarks and analysis. To fill this gap, we introduce ShotFinder, a benchmark that formalizes editing requirements as keyframe-oriented shot descriptions and introduces five types of controllable single-factor constraints: Temporal order, Color, Visual style, Audio, and Resolution. We curate 1,210 high-quality samples from YouTube across 20 thematic categories, using large models for generation with human verification. Based on the benchmark, we propose ShotFinder, a text-driven three-stage retrieval and localization pipeline: (1) query expansion via video imagination, (2) candidate video retrieval with a search engine, and (3) description-guided temporal localization. Experiments on multiple closed-source and open-source models reveal a significant gap to human performance, with clear imbalance across constraints: temporal localization is relatively tractable, while color and visual style remain major challenges. These results reveal that open-domain video shot retrieval is still a critical capability that multimodal large models have yet to overcome.
Authors: Aleksandr Zimin, Yury Polyanskiy, Philippe Rigollet
Abstract: We propose a variational framework that interprets transformer layers as iterations of an optimization algorithm acting on token embeddings. In this view, self-attention implements a gradient step of an interaction energy, while MLP layers correspond to gradient updates of a potential energy. Standard GPT-style transformers emerge as vanilla gradient descent on the resulting composite objective, implemented via Lie--Trotter splitting between these two energy functionals. This perspective enables principled architectural design using classical optimization ideas. As a proof of concept, we introduce a Nesterov-style accelerated transformer that preserves the same attention and MLP oracles. The resulting architecture consistently outperforms a nanoGPT baseline on TinyStories and OpenWebText, demonstrating that optimization-theoretic insights can translate into practical gains.
Authors: Ye Yu, Haibo Jin, Yaoning Yu, Jun Zhuang, Haohan Wang
Abstract: Large audio-language models increasingly operate on raw speech inputs, enabling more seamless integration across domains such as voice assistants, education, and clinical triage. This transition, however, introduces a distinct class of vulnerabilities that remain largely uncharacterized. We examine the security implications of this modality shift by designing a text-to-audio jailbreak that embeds disallowed directives within a narrative-style audio stream. The attack leverages an advanced instruction-following text-to-speech (TTS) model to exploit structural and acoustic properties, thereby circumventing safety mechanisms primarily calibrated for text. When delivered through synthetic speech, the narrative format elicits restricted outputs from state-of-the-art models, including Gemini 2.0 Flash, achieving a 98.26% success rate that substantially exceeds text-only baselines. These results highlight the need for safety frameworks that jointly reason over linguistic and paralinguistic representations, particularly as speech-based interfaces become more prevalent.
Authors: Mikael M{\o}ller H{\o}gsgaard, Chirag Pabbaraju
Abstract: Recent works on language identification and generation have established tight statistical rates at which these tasks can be achieved. These works typically operate under a strong realizability assumption: that the input data is drawn from an unknown distribution necessarily supported on some language in a given collection. In this work, we relax this assumption of realizability entirely, and impose no restrictions on the distribution of the input data. We propose objectives to study both language identification and generation in this more general "agnostic" setup. Across both problems, we obtain novel interesting characterizations and nearly tight rates.
Authors: Ruijie Zhang, Yequan Zhao, Ziyue Liu, Zhengyang Wang, Dongyang Li, Yupeng Su, Sijia Liu, Zheng Zhang
Abstract: The Muon optimizer has demonstrated strong empirical performance in pre-training large language models by performing matrix-level gradient (or momentum) orthogonalization in each layer independently. In this work, we propose TEON, a principled generalization of Muon that extends orthogonalization beyond individual layers by modeling the gradients of a neural network as a structured higher-order tensor. We present TEON's improved convergence guarantee over layer-wise Muon, and further develop a practical instantiation of TEON based on the theoretical analysis with corresponding ablation. We evaluate our approach on two widely adopted architectures: GPT-style models, ranging from 130M to 774M parameters, and LLaMA-style models, ranging from 60M to 1B parameters. Experimental results show that TEON consistently improves training and validation perplexity across model scales and exhibits strong robustness under various approximate SVD schemes.
Authors: Seyed Ahmad Hosseini Miangoleh, Amin Jalal Aghdasian, Farzaneh Abdollahi
Abstract: This paper proposes a novel inverse reinforcement learning framework using a diffusion-based adaptive lookahead planner (IRL-DAL) for autonomous vehicles. Training begins with imitation from an expert finite state machine (FSM) controller to provide a stable initialization. Environment terms are combined with an IRL discriminator signal to align with expert goals. Reinforcement learning (RL) is then performed with a hybrid reward that combines diffuse environmental feedback and targeted IRL rewards. A conditional diffusion model, which acts as a safety supervisor, plans safe paths. It stays in its lane, avoids obstacles, and moves smoothly. Then, a learnable adaptive mask (LAM) improves perception. It shifts visual attention based on vehicle speed and nearby hazards. After FSM-based imitation, the policy is fine-tuned with Proximal Policy Optimization (PPO). Training is run in the Webots simulator with a two-stage curriculum. A 96\% success rate is reached, and collisions are reduced to 0.05 per 1k steps, marking a new benchmark for safe navigation. By applying the proposed approach, the agent not only drives in lane but also handles unsafe conditions at an expert level, increasing robustness.We make our code publicly available.
Authors: MH Farhadi, Ali Rabiee, Sima Ghafoori, Anna Cetera, Andrew Fisher, Reza Abiri
Abstract: Shared autonomy systems require principled methods for inferring user intent and determining appropriate assistance levels. This is a central challenge in human-robot interaction, where systems must be successful while being mindful of user agency. Previous approaches relied on static blending ratios or separated goal inference from assistance arbitration, leading to suboptimal performance in unstructured environments. We introduce BRACE (Bayesian Reinforcement Assistance with Context Encoding), a novel framework that fine-tunes Bayesian intent inference and context-adaptive assistance through an architecture enabling end-to-end gradient flow between intent inference and assistance arbitration. Our pipeline conditions collaborative control policies on environmental context and complete goal probability distributions. We provide analysis showing (1) optimal assistance levels should decrease with goal uncertainty and increase with environmental constraint severity, and (2) integrating belief information into policy learning yields a quadratic expected regret advantage over sequential approaches. We validated our algorithm against SOTA methods (IDA, DQN) using a three-part evaluation progressively isolating distinct challenges of end-effector control: (1) core human-interaction dynamics in a 2D human-in-the-loop cursor task, (2) non-linear dynamics of a robotic arm, and (3) integrated manipulation under goal ambiguity and environmental constraints. We demonstrate improvements over SOTA, achieving 6.3% higher success rates and 41% increased path efficiency, and 36.3% success rate and 87% path efficiency improvement over unassisted control. Our results confirmed that integrated optimization is most beneficial in complex, goal-ambiguous scenarios, and is generalizable across robotic domains requiring goal-directed assistance, advancing the SOTA for adaptive shared autonomy.
Authors: Hongyang Du, Junjie Ye, Xiaoyan Cong, Runhao Li, Jingcheng Ni, Aman Agarwal, Zeqi Zhou, Zekun Li, Randall Balestriero, Yue Wang
Abstract: While recent video diffusion models (VDMs) produce visually impressive results, they fundamentally struggle to maintain 3D structural consistency, often resulting in object deformation or spatial drift. We hypothesize that these failures arise because standard denoising objectives lack explicit incentives for geometric coherence. To address this, we introduce VideoGPA (Video Geometric Preference Alignment), a data-efficient self-supervised framework that leverages a geometry foundation model to automatically derive dense preference signals that guide VDMs via Direct Preference Optimization (DPO). This approach effectively steers the generative distribution toward inherent 3D consistency without requiring human annotations. VideoGPA significantly enhances temporal stability, physical plausibility, and motion coherence using minimal preference pairs, consistently outperforming state-of-the-art baselines in extensive experiments.
Authors: Sanidhya Vijayvargiya, Xuhui Zhou, Akhila Yerukola, Maarten Sap, Graham Neubig
Abstract: AI agents are increasingly being deployed to automate tasks, often based on underspecified user instructions. Making unwarranted assumptions to compensate for the missing information and failing to ask clarifying questions can lead to suboptimal outcomes, safety risks due to tool misuse, and wasted computational resources. In this work, we study the ability of LLM agents to handle underspecified instructions in interactive code generation settings by evaluating proprietary and open-weight models on their performance across three key steps: (a) detecting underspecificity, (b) asking targeted clarification questions, and (c) leveraging the interaction to improve performance in underspecified scenarios. We introduce Ambig-SWE, an underspecified variant of SWE-Bench Verified, specifically designed to evaluate agent behavior under ambiguity and interaction. Our findings reveal that models struggle to distinguish between well-specified and underspecified instructions. However, when models interact for underspecified inputs, they effectively obtain vital information from the user leading to significant improvements in performance, up to 74% over the non-interactive settings, underscoring the value of effective interaction. Our study highlights critical gaps in how current state-of-the-art models handle missing information in complex software engineering tasks and structures the evaluation into distinct steps to enable targeted improvements.
Authors: Tobias Schnabel, Kiran Tomlinson, Adith Swaminathan, Jennifer Neville
Abstract: Despite their many successes, transformer-based large language models (LLMs) continue to struggle with tasks that require complex reasoning over large parts of their input. We argue that these failures arise due to capacity limits on the accurate flow of information within LLMs. To formalize this issue, we introduce the bounded attention prefix oracle (BAPO) model, a new computational framework that models bandwidth constraints on attention heads, the mechanism for internal communication in LLMs. We show that several important reasoning problems like graph reachability require high communication bandwidth for BAPOs to solve; we call these problems BAPO-hard. Our experiments corroborate our theoretical predictions: GPT-4o, Claude, and Gemini succeed on BAPO-easy tasks and fail even on relatively small BAPO-hard tasks. BAPOs also reveal another benefit of chain of thought (CoT): we prove that breaking down a task using CoT can turn any BAPO-hard problem into a BAPO-easy one. Our results offer principled explanations for key LLM failures and suggest directions for architectures and inference methods that mitigate bandwidth limits.
Authors: Danqing Chen, Tobias Ladner, Ahmed Rayen Mhadhbi, Matthias Althoff
Abstract: As large language models become integral to high-stakes applications, ensuring their robustness and fairness is critical. Despite their success, large language models remain vulnerable to adversarial attacks, where small perturbations, such as synonym substitutions, can alter model predictions, posing risks in fairness-critical areas, such as gender bias mitigation, and safety-critical areas, such as toxicity detection. While formal verification has been explored for neural networks, its application to large language models remains limited. This work presents a holistic verification framework to certify the robustness of transformer-based language models, with a focus on ensuring gender fairness and consistent outputs across different gender-related terms. Furthermore, we extend this methodology to toxicity detection, offering formal guarantees that adversarially manipulated toxic inputs are consistently detected and appropriately censored, thereby ensuring the reliability of moderation systems. By formalizing robustness within the embedding space, this work strengthens the reliability of language models in ethical AI deployment and content moderation.
Authors: Safal Shrestha, Minwu Kim, Aadim Nepal, Anubhav Shrestha, Keith Ross
Abstract: Designing effective reasoning-capable LLMs typically requires training using Reinforcement Learning with Verifiable Rewards (RLVR) or distillation with carefully curated Long Chain of Thoughts (CoT), both of which depend heavily on extensive training data. This creates a major challenge when the amount of quality training data is scarce. We propose a sample-efficient, two-stage training strategy to develop reasoning LLMs under limited supervision. In the first stage, we "warm up" the model by distilling Long CoTs from a toy domain, namely, Knights \& Knaves (K\&K) logic puzzles to acquire general reasoning skills. In the second stage, we apply RLVR to the warmed-up model using a limited set of target-domain examples. Our experiments demonstrate that this two-phase approach offers several benefits: $(i)$ the warmup phase alone facilitates generalized reasoning, leading to performance improvements across a range of tasks, including MATH, HumanEval$^{+}$, and MMLU-Pro; $(ii)$ When both the base model and the warmed-up model are RLVR trained on the same small dataset ($\leq100$ examples), the warmed-up model consistently outperforms the base model; $(iii)$ Warming up before RLVR training allows a model to maintain cross-domain generalizability even after training on a specific domain; $(iv)$ Introducing warmup in the pipeline improves not only accuracy but also overall sample efficiency during RLVR training. The results in this paper highlight the promise of warmup for building robust reasoning LLMs in data-scarce environments.
Authors: Xin Shu, Shuai Wang, Ang Li
Abstract: Probabilities of causation (PoCs) are fundamental quantities for counterfactual analysis and personalized decision making. However, existing analytical results are largely confined to binary settings. This paper extends PoCs to multi-valued treatments and outcomes by deriving closed form bounds for a representative family of discrete PoCs within Structural Causal Models, using standard experimental and observational distributions. We introduce the notion of equivalence classes of PoCs, which reduces arbitrary discrete PoCs to this family, and establish a replaceability principle that transfers bounds across value permutations. For the resulting bounds, we prove soundness in all dimensions and empirically verify tightness in low dimensional cases via Balke's linear programming method; we further conjecture that this tightness extends to all dimensions. Simulations indicate that our closed form bounds consistently tighten recent recursive bounds while remaining simpler to compute. Finally, we illustrate the practical relevance of our results through toy examples.
Authors: Fedor Rodionov, Abdelrahman Eldesokey, Michael Birsak, John Femiani, Bernard Ghanem, Peter Wonka
Abstract: We introduce FloorplanQA, a diagnostic benchmark for evaluating spatial reasoning in large-language models (LLMs). FloorplanQA is grounded in structured representations of indoor scenes, such as (e.g., kitchens, living rooms, bedrooms, bathrooms, and others), encoded symbolically in JSON or XML layouts. The benchmark covers core spatial tasks, including distance measurement, visibility, path finding, and object placement within constrained spaces. Our results across a variety of frontier open-source and commercial LLMs reveal that while models may succeed in shallow queries, they often fail to respect physical constraints, preserve spatial coherence, though they remain mostly robust to small spatial perturbations. FloorplanQA uncovers a blind spot in today's LLMs: inconsistent reasoning about indoor layouts. We hope this benchmark inspires new work on language models that can accurately infer and manipulate spatial and geometric properties in practical settings.
Authors: Yunhao Yang, Neel P. Bhatt, Christian Ellis, Samuel Li, Alvaro Velasquez, Zhangyang Wang, Ufuk Topcu
Abstract: Logistics operators, from battlefield coordinators re-routing airlifts ahead of a storm to warehouse managers juggling late trucks, need to make mission-critical decisions. Prevailing methods for logistics planning such as integer programming yield plans that satisfy user-defined logical constraints, assuming an idealized mathematical model of the environment. On the other hand, foundation models lower the intermediate processing barrier by translating natural-language user utterances into executable plans, yet they remain prone to misinterpretations and hallucinations that jeopardize safety and cost. We introduce a Vision-Language Logistics (VLL) agent, built on a neurosymbolic framework that pairs the accessibility of natural-language dialogue with verifiable guarantees on user-objective interpretation. The agent interprets user requests and converts them into structured planning specifications, quantifies the uncertainty of the interpretation, and invokes an interactive clarification loop when the uncertainty exceeds an adaptive threshold. Drawing on a lightweight airlift logistics planning use case as an illustrative case study, we highlight a practical path toward certifiable and user-aligned decision-making for complex logistics. Our lightweight model, fine-tuned on just 100 training samples, surpasses the zero-shot performance of 20x larger models in logistic planning tasks while cutting inference latency by nearly 50%.
Authors: Andrea Asperti, Alberto Naibo, Claudio Sacerdoti Coen
Abstract: Large Language Models (LLMs) have demonstrated impressive capabilities in structured reasoning and symbolic tasks, with coding emerging as a particularly successful application. This progress has naturally motivated efforts to extend these models to mathematics, both in its traditional form, expressed through natural-style mathematical language, and in its formalized counterpart, expressed in a symbolic syntax suitable for automatic verification. Yet, despite apparent parallels between programming and proof construction, advances in formalized mathematics have proven significantly more challenging. This gap raises fundamental questions about the nature of reasoning in current LLM architectures, the role of supervision and feedback, and the extent to which such models maintain an internal notion of computational or deductive state. In this article, we review the current state-of-the-art in mathematical reasoning with LLMs, focusing on recent models and benchmarks. We explore three central issues at the intersection of machine learning and mathematical cognition: (i) the trade-offs between traditional and formalized mathematics as training and evaluation domains; (ii) the structural and methodological reasons why proof synthesis remains more brittle than code generation; and (iii) whether LLMs genuinely represent or merely emulate a notion of evolving logical state. Our goal is not to draw rigid distinctions but to clarify the present boundaries of these systems and outline promising directions for their extension.
Authors: Xuhui Zhou, Jiarui Liu, Akhila Yerukola, Hyunwoo Kim, Maarten Sap
Abstract: Humans intuitively navigate social interactions by simulating unspoken dynamics and reasoning about others' perspectives, even with limited information. In contrast, AI systems struggle to structure and reason about implicit social contexts, as they lack explicit representations for unobserved dynamics such as intentions, beliefs, and evolving social states. In this paper, we introduce the concept of social world models (SWMs) to characterize the complex social dynamics. To operationalize SWMs, we introduce a novel structured social world representation formalism (S3AP), which captures the evolving states, actions, and mental states of agents, addressing the lack of explicit structure in traditional free-text-based inputs. Through comprehensive experiments across five social reasoning benchmarks, we show that S3AP significantly enhances LLM performance-achieving a +51% improvement on FANToM over OpenAI's o1. Our ablations further reveal that these gains are driven by the explicit modeling of hidden mental states, which proves more effective than a wide range of baseline methods. Finally, we introduce an algorithm for social world models using S3AP, which enables AI agents to build models of their interlocutors and predict their next actions and mental states. Empirically, S3AP-enabled social world models yield up to +18% improvement on the SOTOPIA multi-turn social interaction benchmark. Our findings highlight the promise of S3AP as a powerful, general-purpose representation for social world states, enabling the development of more socially-aware systems that better navigate social interactions.
Authors: Chenyang Zhu, Spencer Hong, Jingyu Wu, Kushal Chawla, Charlotte Tang, Youbing Yin, Nathan Wolfe, Erin Babinsky, Daben Liu
Abstract: The advent of complex, interconnected long-horizon LLM systems has made it incredibly tricky to identify where and when these systems break down. Evaluation capabilities that currently exist today are limited in that they often focus on simple metrics, end-to-end outcomes, and are dependent on the perspectives of humans. In order to match the increasing complexity of these many component systems, evaluation frameworks must also be able to reason, probe, iterate, and understand the nuanced logic passing through these systems. In this paper, we present RAFFLES, an offline evaluation architecture that incorporates iterative reasoning. Specifically, RAFFLES operates as an iterative, multi-component pipeline, using a central Judge to systematically identify faults and a set of specialized Evaluators to assess the quality of the candidate faults as well as rationales of the Judge. We evaluated RAFFLES with several benchmarks - the Who&When dataset to identify step-level faults in multi-agent systems and the ReasonEval datasets to diagnose step-level mathematical reasoning errors. RAFFLES outperforms strong baselines, achieving an accuracy of over 20% and 50% on the Who&When Hand-Crafted and Algorithmically-Generated datasets, and over 80% on the ReasonEval datasets. These results demonstrate a key step towards introducing automated fault detection for autonomous systems over labor-intensive manual review.
Authors: Binghan Wu, Shoufeng Wang, Yunxin Liu, Ya-Qin Zhang, Joseph Sifakis, Ye Ouyang
Abstract: The evolution toward Level 4 (L4) Autonomous Networks (AN) represents a strategic inflection point in telecommunications, where networks must transcend reactive automation to achieve genuine cognitive capabilities--fulfilling TM Forum's vision of self-configuring, self-healing, and self-optimizing systems that deliver zero-wait, zero-touch, and zero-fault services. This work bridges the gap between architectural theory and operational reality by implementing Joseph Sifakis's AN Agent reference architecture in a functional cognitive system, deploying coordinated proactive-reactive runtimes driven by hybrid knowledge representation. Through an empirical case study of a Radio Access Network (RAN) Link Adaptation (LA) Agent, we validate this framework's transformative potential: demonstrating sub-10 ms real-time control in 5G NR sub-6 GHz while achieving 4% higher downlink throughput than Outer Loop Link Adaptation (OLLA) algorithms and 85% Block Error Rate (BLER) reduction for ultra-reliable services through dynamic Modulation and Coding Scheme (MCS) optimization. These improvements confirm the architecture's viability in overcoming traditional autonomy barriers and advancing critical L4-enabling capabilities toward next-generation objectives.
Authors: Yiqun Yao, Naitong Yu, Xiang Li, Xin Jiang, Xuezhi Fang, Wenjia Ma, Xuying Meng, Jing Li, Aixin Sun, Yequan Wang
Abstract: We introduce EgoMem, the first lifelong memory agent tailored for full-duplex models that process real-time omnimodal streams. EgoMem enables real-time models to recognize multiple users directly from raw audiovisual streams, to provide personalized response, and to maintain long-term knowledge of users' facts, preferences, and social relationships extracted from audiovisual history. EgoMem operates with three asynchronous processes: (i) a retrieval process that dynamically identifies user via face and voice, and gathers relevant context from a long-term memory; (ii) an omnimodal dialog process that generates personalized audio responses based on the retrieved context; and (iii) a memory management process that automatically detects dialog boundaries from omnimodal streams, and extracts necessary information to update the long-term memory. Unlike existing memory agents for LLMs, EgoMem relies entirely on raw audiovisual streams, making it especially suitable for lifelong, real-time, and embodied scenarios. Experimental results demonstrate that EgoMem's retrieval and memory management modules achieve over 95% accuracy on the test set. When integrated with a fine-tuned RoboEgo omnimodal chatbot, the system achieves fact-consistency scores above 87% in real-time personalized dialogs, establishing a strong baseline for future research.
Authors: Ankur Samanta, Akshayaa Magesh, Runzhe Wu, Ayush Jain, Youliang Yu, Daniel Jiang, Boris Vidolov, Paul Sajda, Yonathan Efroni, Kaveh Hassani
Abstract: Self-improvement, where models improve beyond their current performance without external supervision, remains a challenge. The core difficulty is sourcing a training signal stronger than what the model itself can currently produce. Majority voting has been shown to provide such a signal by aggregating over multiple samples, helping mitigate some of the inconsistencies in LM reasoning. In this work, we show that multi-agent debate--where models collaborate and exchange reasoning over multiple rounds--provides an even richer signal than single-round majority voting. We introduce Multi-Agent Consensus Alignment (MACA), which uses reinforcement learning (RL) to post-train models to effectively utilize multi-agent debate. We find that preference learning over full reasoning traces, learning to differentiate between majority and minority reasoning, is more effective than binary consensus rewards or SFT-based approaches for leveraging these debate signals. This produces three key improvements: models are (1) better at utilizing the multi-agent debate setting (+26.87% on MATH), (2) individually more accurate (+21.51% on MathQA), and (3) more self-consistent (+27.6% on GSM8K). We also see strong generalization to unseen benchmarks (+16.3% on GPQA, +11.6% on CommonsenseQA).
Authors: Debarpan Bhattacharya, Apoorva Kulkarni, Sriram Ganapathy
Abstract: The accurate trust assessment of multimodal large language models (MLLMs) generated predictions, which can enable selective prediction and improve user confidence, is challenging due to the diverse multi-modal input paradigms. We propose Functionally Equivalent Sampling for Trust Assessment (FESTA), a multimodal input sampling technique for MLLMs, that generates an uncertainty measure based on the equivalent and complementary input samplings. The proposed task-preserving sampling approach for uncertainty quantification expands the input space to probe the consistency (through equivalent samples) and sensitivity (through complementary samples) of the model. FESTA uses only input-output access of the model (black-box), and does not require ground truth (unsupervised). The experiments are conducted with various off-the-shelf multi-modal LLMs, on both visual and audio reasoning tasks. The proposed FESTA uncertainty estimate achieves significant improvement (33.3% relative improvement for vision-LLMs and 29.6% relative improvement for audio-LLMs) in selective prediction performance, based on area-under-receiver-operating-characteristic curve (AUROC) metric in detecting mispredictions. The code implementation is open-sourced.
Authors: Jiyuan Pei, Yi Mei, Jialin Liu, Mengjie Zhang, Xin Yao
Abstract: Recent neural solvers have demonstrated promising performance in learning to solve routing problems. However, existing studies are primarily based on one-off training on one or a set of predefined problem distributions and scales, i.e., tasks. When a new task arises, they typically rely on either zero-shot generalization, which may be poor due to the discrepancies between the new task and the training task(s), or fine-tuning the pretrained solver on the new task, which possibly leads to catastrophic forgetting of knowledge acquired from previous tasks. This paper explores a novel lifelong learning paradigm for neural VRP solvers, where multiple tasks with diverse distributions and scales arise sequentially over time. Solvers are required to effectively and efficiently learn to solve new tasks while maintaining their performance on previously learned tasks. Consequently, a novel framework called Lifelong Learning Router with Behavior Consolidation (LLR-BC) is proposed. LLR-BC consolidates prior knowledge effectively by aligning behaviors of the solver trained on a new task with the buffered ones in a decision-seeking way. To encourage more focus on crucial experiences, LLR-BC assigns greater consolidated weights to decisions with lower confidence. Extensive experiments on capacitated vehicle routing problems and traveling salesman problems demonstrate LLR-BC's effectiveness in training high-performance neural solvers in a lifelong learning setting, addressing the catastrophic forgetting issue, maintaining their plasticity, and improving zero-shot generalization ability.
Authors: Zhimin Wang, Duo Wu, Shaokang He, Jinghe Wang, Linjia Kang, Jing Yu, Kai Zhu, Jiawei Li, Zhi Wang
Abstract: Effective real-world multi-agent collaboration requires not only accurate planning but also the ability to reason about collaborators' intents--a crucial capability for avoiding miscoordination and redundant communication under partial observable environments. Due to their strong planning and reasoning capabilities, large language models (LLMs) have emerged as promising autonomous agents for collaborative task solving. However, existing collaboration frameworks for LLMs overlook their reasoning potential for dynamic intent inference, and thus produce inconsistent plans and redundant communication, reducing collaboration efficiency. To bridge this gap, we propose CoBel-World, a novel framework that equips LLM agents with a Collaborative Belief World--an internal representation jointly modeling the physical environment and collaborators' mental states. CoBel-World enables agents to parse external open-world knowledge into structured beliefs via a symbolic belief representation module, and perform zero-shot Bayesian-style belief updates through LLM reasoning. This allows agents to proactively detect potential miscoordination (e.g., conflicting plans) and communicate adaptively. Evaluated on challenging embodied benchmarks (i.e., TDW-MAT and C-WAH), CoBel-World significantly reduces communication costs by 64-79% and improves task completion efficiency by 4-28% compared to the strongest baseline. Our results show that explicit, intent-aware belief modeling is essential for efficient and human-like collaboration in LLM-based multi-agent systems.
Authors: Yihang Chen, Yuanhao Ban, Yunqi Hong, Cho-Jui Hsieh
Abstract: Despite the success of Reinforcement Learning from Human Feedback (RLHF) in language reasoning, its application to autoregressive Text-to-Image (T2I) generation is often constrained by the limited availability of human preference data. This paper explores how an autoregressive T2I model can learn from internal signals without relying on external rewards or labeled data. Contrary to recent findings in math and code reasoning, we show that minimizing self-certainty, rather than maximizing it, improves image generation. We observe that autoregressive T2I models with higher certainty are likely to generate simple and uniform images, which are less aligned with human preferences, and models with lower certainty are likely to generate vivid images rich in detail. Based on this observation, we propose IRIS(Intrinsic Reward Image Synthesis), the first framework to improve autoregressive T2I models with reinforcement learning using only an intrinsic reward. Empirical results demonstrate that applying IRIS to autoregressive T2I models achieves performance superior to those trained by individual external rewards, and matching those trained by ensemble external rewards. IRIS also incentivizes the emergence of nuanced CoT reasoning for high-quality image generation.
Authors: Yubin Kim, Taehan Kim, Eugene Park, Chunjong Park, Cynthia Breazeal, Daniel McDuff, Hae Won Park
Abstract: We present InvThink, a simple yet powerful approach that gives language models the capability of inverse thinking: reasoning through failure modes before generating responses. Unlike existing safety alignment methods that optimize directly for safe response, InvThink instructs models to 1) enumerate potential harms, 2) analyze their consequences, and 3) generate safe outputs that proactively avoid these risks. Our paper reveals three key findings: (i) InvThink demonstrates significantly improved safety reasoning as model size scales, compared to existing safety methods. (ii) InvThink mitigates safety tax; by training models to systematically consider failure modes, it preserves general reasoning capabilities on standard benchmarks. (iii) beyond general safety tasks, InvThink excels in high-stakes domains including external-facing applications (medicine, finance, law) and agentic risk scenarios (blackmail, murder), achieving up to 17.8% reduction in harmful responses compared to baseline methods like SafetyPrompt. We further equip InvThink with supervised fine-tuning, and reinforcement learning across three LLM families. These results suggest that InvThink provides a scalable and generalizable path toward safer, more capable language models.
Authors: Hao Zeng, Jianguo Huang, Bingyi Jing, Hongxin Wei, Bo An
Abstract: Large reasoning models (LRMs) have achieved remarkable progress in complex problem-solving tasks. Despite this success, LRMs typically suffer from high computational costs during deployment, highlighting a need for efficient inference. A practical direction of efficiency improvement is to switch the LRM between thinking and non-thinking modes dynamically. However, such approaches often introduce additional reasoning errors and lack statistical guarantees for the performance loss, which are critical for high-stakes applications. In this work, we propose Probably Approximately Correct (PAC) reasoning that controls the performance loss under the user-specified tolerance. Specifically, we construct an upper confidence bound on the performance loss and determine a threshold for switching to the non-thinking model. Theoretically, using the threshold to switch between the thinking and non-thinking modes ensures bounded performance loss in a distribution-free manner. Our comprehensive experiments on reasoning benchmarks show that the proposed method can save computational budgets and control the user-specified performance loss.
Authors: Siyuan Lu, Zechuan Wang, Hongxuan Zhang, Qintong Wu, Leilei Gan, Chenyi Zhuang, Jinjie Gu, Tao Lin
Abstract: Large Language Model (LLM) agents show great promise for complex, multi-turn tool-use tasks, but their development is often hampered by the extreme scarcity of high-quality training data. Supervised fine-tuning (SFT) on synthetic data leads to overfitting, whereas standard reinforcement learning (RL) struggles with a critical cold-start problem and training instability. To address these challenges, we introduce $\textbf{Environment Tuning}$, a novel training paradigm that enables agents to learn complex behaviors directly from problem instances without relying on pre-collected expert trajectories. $\textbf{Environment Tuning}$ orchestrates this learning process through a structured curriculum, actionable environment augmentation that provides corrective feedback, and fine-grained progress rewards to ensure stable and efficient exploration. Using only 400 problem instances from Berkeley Function-Calling Leaderboard (BFCL) benchmark, our method not only achieves competitive in-distribution performance against strong baselines but also demonstrates superior out-of-distribution generalization, overcoming the performance collapse common to SFT-based approaches. Our work presents a paradigm shift from supervised fine-tuning on static trajectories to dynamic, environment-based exploration, paving the way for training more robust and data-efficient agents. The code is available at https://github.com/inclusionAI/AWorld-RL/tree/main/EnvTuning.
URLs: https://github.com/inclusionAI/AWorld-RL/tree/main/EnvTuning.
Authors: Daoyu Wang, Mingyue Cheng, Shuo Yu, Zirui Liu, Ze Guo, Xin Li, Qi Liu
Abstract: Understanding and reasoning on the large-scale scientific literature is a crucial touchstone for large language model (LLM) based agents. However, existing works are mainly restricted to tool-free tasks within single papers, largely due to the lack of a benchmark that evaluates cross-paper reasoning and multi-tool orchestration in authentic research scenarios. In this work, we propose PaperArena, a benchmark to evaluate LLM-based agents on questions that require integrating information across multiple papers with the assistance of external tools. Given a research question, agents should formulate a reasoning plan, interact with multiple papers, and invoke appropriate tools to produce a well-grounded answer. To support standardized evaluation, we provide a platform for agent execution, offering a modular tool environment including multimodal parsing, context retrieval, and programmatic computation. Experiments reveal that even the leading LLM powering a well-established agentic workflow achieves merely 38.78% average accuracy, while on the hard subset, accuracy drops to only 18.47%. We also analyze reasoning traces and diagnose agent behavior, providing the community with insights to develop and evaluate more capable scientific agents.
Authors: Roham Koohestani, Ziyou Li, Anton Podkopaev, Maliheh Izadi
Abstract: This paper studies standard controller architectures for agentic AI and derives automata-theoretic models of their interaction behavior via trace semantics and abstraction. We model an agent implementation as a finite control program augmented with explicit memory primitives (bounded buffers, a call stack, or read/write external memory) and a stochastic policy component (e.g., an LLM) that selects among architecturally permitted actions. Instead of equating the concrete agent with a deterministic acceptor, we treat the agent-environment closed loop as inducing a probability distribution over finite interaction traces. Given an abstraction function $\Abs$ from concrete configurations to a finite abstract state space, we obtain a probabilistic trace language and an abstract probabilistic transition model $M_{\Abs}$ suitable for probabilistic model checking. Imposing explicit, framework-auditable restrictions on memory access and control flow, we prove that the support of the resulting trace language is regular for bounded-memory controllers, context-free for strict call-return controllers, and recursively enumerable for controllers equipped with unbounded read/write memory. These correspondences allow the reuse of existing verification methods for finite-state and pushdown systems, and they delineate precisely when undecidability barriers arise. The probabilistic semantics leads to quantitative analyses such as: what is the probability of entering an unsafe abstract region, and how can we bound this probability in the presence of environment nondeterminism.
Authors: Willem Fourie
Abstract: Instrumental goals such as resource acquisition, power-seeking, and self-preservation are key to contemporary AI alignment research, yet the phenomenon's ontology remains under-theorised. This article develops an ontological account of instrumental goals and draws out governance-relevant distinctions for advanced AI systems. After systematising the dominant alignment literature on instrumental goals we offer an exploratory Aristotelian framework that treats advanced AI systems as complex artefacts whose ends are externally imposed through design, training and deployment. On a structural reading, Aristotle's notion of hypothetical necessity explains why, given an imposed end pursued over extended horizons in particular environments, certain enabling conditions become conditionally required, thereby yielding robust instrumental tendencies. On a contingent reading, accidental causation and chance-like intersections among training regimes, user inputs, infrastructure and deployment contexts can generate instrumental-goal-like behaviours not entailed by the imposed end-structure. This dual-aspect ontology motivates for governance and management approaches that treat instrumental goals as features of advanced AI systems to be managed rather than anomalies eliminable by technical interventions.
Authors: Qianli Shen, Daoyuan Chen, Yilun Huang, Zhenqing Ling, Yaliang Li, Bolin Ding, Jingren Zhou
Abstract: Reinforcement finetuning (RFT) is a key technique for aligning Large Language Models (LLMs) with human preferences and enhancing reasoning, yet its effectiveness is highly sensitive to which tasks are explored during training. Uniform task sampling is inefficient, wasting computation on tasks that are either trivial or unsolvable, while existing task selection methods often suffer from high rollout costs, poor adaptivity, or incomplete evidence. We introduce BOTS, a unified framework for Bayesian Online Task Selection in LLM reinforcement finetuning. Grounded in Bayesian inference, BOTS adaptively maintains posterior estimates of task difficulty as the model evolves. It jointly incorporates explicit evidence from direct evaluations of selected tasks and implicit evidence inferred from these evaluations for unselected tasks, with Thompson sampling ensuring a principled balance between exploration and exploitation for task selection. To make implicit evidence practical, we instantiate it with an ultra-light interpolation-based plug-in that estimates difficulties of tasks without extra rollouts, adding negligible overhead. Empirically, across diverse domains and LLM scales, BOTS consistently improves data efficiency and performance over baselines and ablations, providing a practical and extensible solution for dynamic task selection in RFT. Code is available at https://github.com/agentscope-ai/Trinity-RFT/tree/main/examples/bots.
URLs: https://github.com/agentscope-ai/Trinity-RFT/tree/main/examples/bots.
Authors: Lingyue Fu, Xin Ding, Linyue Pan, Yaoming Zhu, Shao Zhang, Lin Qiu, Weiwen Liu, Weinan Zhang, Xuezhi Cao, Xunliang Cai, Jiaxin Ding, Yong Yu
Abstract: Current evaluation for Large Language Model (LLM) code agents predominantly focus on generating functional code in single-turn scenarios, which fails to evaluate the agent's capability for continuous code optimization and multi-turn iterative development. To bridge this gap, we introduce CATArena, a framework designed to evaluate the evolutionary capabilities of code agents via iterative tournaments. Agents engage in multi-turn tournaments and continuously refine their code through self-reflection and peer-learning based on comprehensive execution feedback. For evaluation, we propose a dual-metric system to decouple static generation proficiency from evolutionary potential. Extensive experiments reveal that an agent's evolutionary potential is not strictly correlated with its initial proficiency. Our analysis further reveals that current agents struggle to concurrently leverage both peer-learning and self-reflection for effective performance gains. Furthermore, the results validate CATArena's high extensibility and resistance to variance tasks, establishing it as a continuous and reliable standard for assessing the evolutionary capability of LLM code agents.
Authors: Jianwen Sun, Yukang Feng, Yifan Chang, Chuanhao Li, Zizhen Li, Jiaxin Ai, Fanrui Zhang, Yu Dai, Kaipeng Zhang
Abstract: A fundamental bottleneck in human-AI collaboration is the ``intention expression gap," the difficulty for humans to effectively convey complex, high-dimensional thoughts to AI. This challenge often traps users in inefficient trial-and-error loops and is exacerbated by the diverse expertise levels of users. We reframe this problem from passive instruction following to a Socratic collaboration paradigm, proposing an agent that actively probes for information to resolve its uncertainty about user intent. we name the proposed agent Nous, trained to acquire proficiency in this inquiry policy. The core mechanism of Nous is a training framework grounded in the first principles of information theory. Within this framework, we define the information gain from dialogue as an intrinsic reward signal, which is fundamentally equivalent to the reduction of Shannon entropy over a structured task space. This reward design enables us to avoid reliance on costly human preference annotations or external reward models. To validate our framework, we develop an automated simulation pipeline to generate a large-scale, preference-based dataset for the challenging task of scientific diagram generation. Comprehensive experiments, including ablations, subjective and objective evaluations, and tests across user expertise levels, demonstrate the effectiveness of our proposed framework. Nous achieves leading efficiency and output quality, while remaining robust to varying user expertise. In conclusion, our research provides a systematic methodology and a new perspective for addressing the issue of ambiguous intentions in complex human-machine collaboration.
Authors: Han Yu, Xiaojuan Zhao, Aiping Li, Kai Chen, Ziniu Liu, Zhichao Peng
Abstract: Graph neural networks (GNNs) can effectively model structural information of graphs, making them widely used in knowledge graph (KG) reasoning. However, existing studies on the expressive power of GNNs mainly focuses on simple single-relation graphs, and there is still insufficient discussion on the power of GNN to express logical rules in KGs. How to enhance the logical expressive power of GNNs is still a key issue. Motivated by this, we propose Path-Neighbor enhanced GNN (PN-GNN), a method to enhance the logical expressive power of GNN by aggregating node-neighbor embeddings on the reasoning path. First, we analyze the logical expressive power of existing GNN-based methods and point out the shortcomings of the expressive power of these methods. Then, we theoretically investigate the logical expressive power of PN-GNN, showing that it not only has strictly stronger expressive power than C-GNN but also that its $(k+1)$-hop logical expressiveness is strictly superior to that of $k$-hop. Finally, we evaluate the logical expressive power of PN-GNN on six synthetic datasets and two real-world datasets. Both theoretical analysis and extensive experiments confirm that PN-GNN enhances the expressive power of logical rules without compromising generalization, as evidenced by its competitive performance in KG reasoning tasks.
Authors: Yongxian Wei, Yilin Zhao, Li Shen, Xinrui Chen, Runxi Cheng, Sinan Du, Hao Yu, Xiaohan Wang, Gang Liu, Jiahong Yan, Chun Yuan, Dian Li
Abstract: Data synthesis for training large reasoning models offers a scalable alternative to limited, human-curated datasets, enabling the creation of high-quality data. However, existing approaches face several challenges: (i) indiscriminate generation that ignores the solver's ability and yields low-value problems, or reliance on complex data pipelines to balance problem difficulty; and (ii) a lack of reasoning in problem generation, leading to shallow problem variants. In this paper, we develop a problem generator that reasons explicitly to plan problem directions before synthesis and adapts difficulty to the solver's ability. Specifically, we construct related problem pairs and augment them with intermediate problem-design CoT produced by a reasoning model. These data are used to bootstrap problem-design strategies in the generator. Then, we treat the solver's feedback on synthetic problems as a reward signal, enabling the generator to calibrate difficulty and produce complementary problems near the edge of the solver's competence. Extensive experiments on 10 mathematical and general reasoning benchmarks show that our proposed framework achieves a cumulative average improvement of 3.4%, demonstrating robust generalization across both language and vision-language models.
Authors: Xinhang Li, Jingbo Zhou, Pengfei Luo, Yixiong Xiao, Tong Xu
Abstract: Recent advances in multimodal large language models (MLLMs) highlight the need for benchmarks that rigorously evaluate structured chart comprehension. Chart grounding refers to the bidirectional alignment between a chart's visual appearance and its structured semantics. This task requires models to produce a symbolic specification that faithfully captures the chart's visual and structural intent, while also recovering the underlying tabular data with precise values and relationships. Chart grounding directly reflects a model's capabilities in numerical reasoning, multimodal alignment, and structural reconstruction, and has several important real-world applications. Existing benchmarks, constrained by narrow chart diversity, isolated tasks, and incomplete evaluation frameworks, fail to holistically assess grounding. To address this, we propose ChartAnchor, a comprehensive benchmark of 8k+ chart-table-code triples spanning 30 chart types drawn from diverse real-world and augmented sources. ChartAnchor introduces two complementary tasks: chart-to-code generation and controlled chart-to-table reconstruction, enabling cross-validation of visual and numerical fidelity. A multi-level evaluation framework integrates semantic validation, stylistic analysis, and perceptual metrics to assess both structural and content-level correctness. Extensive experiments on MLLMs reveal critical limitations in numerical precision and code synthesis, emphasizing the need for structured reasoning beyond surface-level perception. By unifying symbolic and data-driven grounding, ChartAnchor establishes a rigorous foundation for chart grounding, offering meaningful insights for advancing MLLMs in scientific, financial, and industrial domains.
Authors: Haoyu Jiang, Fanjie Zeng, Boan Qu, Xiaojie Lin, Wei Zhong
Abstract: In the global drive toward carbon neutrality, deeply coordinated smart energy systems underpin industrial transformation. However, the interdisciplinary, fragmented, and fast-evolving expertise in this domain prevents general-purpose LLMs, which lack domain knowledge and physical-constraint awareness, from delivering precise engineering-aligned inference and generation. To address these challenges, we introduce Helios, a large language model tailored to the smart energy domain, together with a comprehensive suite of resources to advance LLM research in this field. Specifically, we develop Enersys, a multi-agent collaborative framework for end-to-end dataset construction, through which we produce: (1) a smart energy knowledge base, EnerBase, to enrich the model's foundational expertise; (2) an instruction fine-tuning dataset, EnerInstruct, to strengthen performance on domain-specific downstream tasks; and (3) an RLHF dataset, EnerReinforce, to align the model with human preferences and industry standards. Leveraging these resources, Helios undergoes large-scale pretraining, SFT, and RLHF. We also release EnerBench, a benchmark for evaluating LLMs in smart energy scenarios, and demonstrate that our approach significantly enhances domain knowledge mastery, task execution accuracy, and alignment with human preferences.
Authors: Yinzhi Zhao, Ming Wang, Shi Feng, Xiaocui Yang, Daling Wang, Yifei Zhang
Abstract: Large language models (LLMs) have achieved impressive performance across natural language tasks and are increasingly deployed in real-world applications. Despite extensive safety alignment efforts, recent studies show that such alignment is often shallow and remains vulnerable to jailbreak attacks. Existing defense mechanisms, including decoding-based constraints and post-hoc content detectors, struggle against sophisticated jailbreaks, often intervening robust detection or excessively degrading model utility. In this work, we examine the decoding process of LLMs and make a key observation: even when successfully jailbroken, models internally exhibit latent safety-related signals during generation. However, these signals are overridden by the model's drive for fluent continuation, preventing timely self-correction or refusal. Building on this observation, we propose a simple yet effective approach that explicitly surfaces and leverages these latent safety signals for early detection of unsafe content during decoding. Experiments across diverse jailbreak attacks demonstrate that our approach significantly enhances safety, while maintaining low over-refusal rates on benign inputs and preserving response quality. Our results suggest that activating intrinsic safety-awareness during decoding offers a promising and complementary direction for defending against jailbreak attacks. Code is available at: https://github.com/zyz13590/SafeProbing.
Authors: Jiayi Yuan, Jonathan N\"other, Natasha Jaques, Goran Radanovi\'c
Abstract: While recent automated red-teaming methods show promise for systematically exposing model vulnerabilities, most existing approaches rely on human-specified workflows. This dependence on manually designed workflows suffers from human biases and makes exploring the broader design space expensive. We introduce AgenticRed, an automated pipeline that leverages LLMs' in-context learning to iteratively design and refine red-teaming systems without human intervention. Rather than optimizing attacker policies within predefined structures, AgenticRed treats red-teaming as a system design problem. Inspired by methods like Meta Agent Search, we develop a novel procedure for evolving agentic systems using evolutionary selection, and apply it to the problem of automatic red-teaming. Red-teaming systems designed by AgenticRed consistently outperform state-of-the-art approaches, achieving 96% attack success rate (ASR) on Llama-2-7B (36% improvement) and 98% on Llama-3-8B on HarmBench. Our approach exhibits strong transferability to proprietary models, achieving 100% ASR on GPT-3.5-Turbo and GPT-4o, and 60% on Claude-Sonnet-3.5 (24% improvement). This work highlights automated system design as a powerful paradigm for AI safety evaluation that can keep pace with rapidly evolving models.
Authors: Kiana Jafari, Paul Ulrich Nikolaus Rust, Duncan Eddy, Robbie Fraser, Nina Vasan, Darja Djordjevic, Akanksha Dadlani, Max Lamparth, Eugenia Kim, Mykel Kochenderfer
Abstract: Learning from human feedback~(LHF) assumes that expert judgments, appropriately aggregated, yield valid ground truth for training and evaluating AI systems. We tested this assumption in mental health, where high safety stakes make expert consensus essential. Three certified psychiatrists independently evaluated LLM-generated responses using a calibrated rubric. Despite similar training and shared instructions, inter-rater reliability was consistently poor ($ICC$ $0.087$--$0.295$), falling below thresholds considered acceptable for consequential assessment. Disagreement was highest on the most safety-critical items. Suicide and self-harm responses produced greater divergence than any other category, and was systematic rather than random. One factor yielded negative reliability (Krippendorff's $\alpha = -0.203$), indicating structured disagreement worse than chance. Qualitative interviews revealed that disagreement reflects coherent but incompatible individual clinical frameworks, safety-first, engagement-centered, and culturally-informed orientations, rather than measurement error. By demonstrating that experts rely on holistic risk heuristics rather than granular factor discrimination, these findings suggest that aggregated labels function as arithmetic compromises that effectively erase grounded professional philosophies. Our results characterize expert disagreement in safety-critical AI as a sociotechnical phenomenon where professional experience introduces sophisticated layers of principled divergence. We discuss implications for reward modeling, safety classification, and evaluation benchmarks, recommending that practitioners shift from consensus-based aggregation to alignment methods that preserve and learn from expert disagreement.
Authors: Hongzhu Yi, Xinming Wang, Zhenghao zhang, Tianyu Zong, Yuanxiang Wang, Jun Xie, Tao Yu, Haopeng Jin, Kaixin Xu, Feng Chen, Jiahuan Chen, Yujia Yang, Zhenyu Guan, Bingkang Shi, Jungang Xu
Abstract: Within the domain of large language models, reinforcement fine-tuning algorithms necessitate the generation of a complete reasoning trajectory beginning from the input query, which incurs significant computational overhead during the rollout phase of training. To address this issue, we analyze the impact of different segments of the reasoning path on the correctness of the final result and, based on these insights, propose Reinforcement Fine-Tuning with Partial Reasoning Optimization (RPO), a plug-and-play reinforcement fine-tuning algorithm. Unlike traditional reinforcement fine-tuning algorithms that generate full reasoning paths, RPO trains the model by generating suffixes of the reasoning path using experience cache. During the rollout phase of training, RPO reduces token generation in this phase by approximately 95%, greatly lowering the theoretical time overhead. Compared with full-path reinforcement fine-tuning algorithms, RPO reduces the training time of the 1.5B model by 90% and the 7B model by 72%. At the same time, it can be integrated with typical algorithms such as GRPO and DAPO, enabling them to achieve training acceleration while maintaining performance comparable to the original algorithms. Our code is open-sourced at https://github.com/yhz5613813/RPO.
Authors: Shuangshuang Ying, Zheyu Wang, Yunjian Peng, Jin Chen, Yuhao Wu, Hongbin Lin, Dingyu He, Siyi Liu, Gengchen Yu, YinZhu Piao, Yuchen Wu, Xin Gui, Zhongyuan Peng, Xin Li, Xeron Du, Libo Qin, YiXin Cao, Ge Zhang, Stephen Huang
Abstract: Despite strong performance on existing benchmarks, it remains unclear whether large language models can reason over genuinely novel scientific information. Most evaluations score end-to-end RAG pipelines, where reasoning is confounded with retrieval and toolchain choices, and the signal is further contaminated by parametric memorization and open-web volatility. We introduce DeR2, a controlled deep-research sandbox that isolates document-grounded reasoning while preserving core difficulties of deep search: multi-step synthesis, denoising, and evidence-based conclusion making. DeR2 decouples evidence access from reasoning via four regimes--Instruction-only, Concepts (gold concepts without documents), Related-only (only relevant documents), and Full-set (relevant documents plus topically related distractors)--yielding interpretable regime gaps that operationalize retrieval loss vs. reasoning loss and enable fine-grained error attribution. To prevent parametric leakage, we apply a two-phase validation that requires parametric failure without evidence while ensuring oracle-concept solvability. To ensure reproducibility, each instance provides a frozen document library (drawn from 2023-2025 theoretical papers) with expert-annotated concepts and validated rationales. Experiments across a diverse set of state-of-the-art foundation models reveal substantial variation and significant headroom: some models exhibit mode-switch fragility, performing worse with the Full-set than with Instruction-only, while others show structural concept misuse, correctly naming concepts but failing to execute them as procedures.
Authors: Bingzhi Zhang, Peng Xu, Xiaohui Chen, Quntao Zhuang
Abstract: Deep generative models are key-enabling technology to computer vision, text generation, and large language models. Denoising diffusion probabilistic models (DDPMs) have recently gained much attention due to their ability to generate diverse and high-quality samples in many computer vision tasks, as well as to incorporate flexible model architectures and a relatively simple training scheme. Quantum generative models, empowered by entanglement and superposition, have brought new insight to learning classical and quantum data. Inspired by the classical counterpart, we propose the quantum denoising diffusion probabilistic model (QuDDPM) to enable efficiently trainable generative learning of quantum data. QuDDPM adopts sufficient layers of circuits to guarantee expressivity, while it introduces multiple intermediate training tasks as interpolation between the target distribution and noise to avoid barren plateau and guarantee efficient training. We provide bounds on the learning error and demonstrate QuDDPM's capability in learning correlated quantum noise model, quantum many-body phases, and topological structure of quantum data. The results provide a paradigm for versatile and efficient quantum generative learning.
Authors: Shahid Alam, Zeynep Altiparmak
Abstract: With the rise of complex cyber devices Cyber Forensics (CF) is facing many new challenges. For example, there are dozens of systems running on smartphones, each with more than millions of downloadable applications. Sifting through this large amount of data and making sense requires new techniques, such as from the field of Artificial Intelligence (AI). To apply these techniques successfully in CF, we need to justify and explain the results to the stakeholders of CF, such as forensic analysts and members of the court, for them to make an informed decision. If we want to apply AI successfully in CF, there is a need to develop trust in AI systems. Some other factors in accepting the use of AI in CF are to make AI authentic, interpretable, understandable, and interactive. This way, AI systems will be more acceptable to the public and ensure alignment with legal standards. An explainable AI (XAI) system can play this role in CF, and we call such a system XAI-CF. XAI-CF is indispensable and is still in its infancy. In this paper, we explore and make a case for the significance and advantages of XAI-CF. We strongly emphasize the need to build a successful and practical XAI-CF system and discuss some of the main requirements and prerequisites of such a system. We present a formal definition of the terms CF and XAI-CF and a comprehensive literature review of previous works that apply and utilize XAI to build and increase trust in CF. We discuss some challenges facing XAI-CF. We also provide some concrete solutions to these challenges. We identify key insights and future research directions for building XAI applications for CF. This paper is an effort to explore and familiarize the readers with the role of XAI applications in CF, and we believe that our work provides a promising basis for future researchers interested in XAI-CF.
Authors: Jaeseung Heo, Moonjeong Park, Dongwoo Kim
Abstract: Label smoothing is a widely studied regularization technique in machine learning. However, its potential for node classification in graph-structured data, spanning homophilic to heterophilic graphs, remains largely unexplored. We introduce posterior label smoothing, a novel method for transductive node classification that derives soft labels from a posterior distribution conditioned on neighborhood labels. The likelihood and prior distributions are estimated from the global statistics of the graph structure, allowing our approach to adapt naturally to various graph properties. We evaluate our method on 10 benchmark datasets using eight baseline models, demonstrating consistent improvements in classification accuracy. The following analysis demonstrates that soft labels mitigate overfitting during training, leading to better generalization performance, and that pseudo-labeling effectively refines the global label statistics of the graph. Our code is available at https://github.com/ml-postech/PosteL.
Authors: Anindya Bijoy Das, Shahnewaz Karim Sakib
Abstract: Large Language Model (LLM)-based recommendation systems excel in delivering comprehensive suggestions by deeply analyzing content and user behavior. However, they often inherit biases from skewed training data, favoring mainstream content while underrepresenting diverse or non-traditional options. This study explores the interplay between bias and LLM-based recommendation systems, focusing on music, song, and book recommendations across diverse demographic and cultural groups. This paper analyzes bias in LLM-based recommendation systems across multiple models (GPT, LLaMA, and Gemini), revealing its deep and pervasive impact on outcomes. Intersecting identities and contextual factors, like socioeconomic status, further amplify biases, complicating fair recommendations across diverse groups. Our findings reveal that bias in these systems is deeply ingrained, yet even simple interventions like prompt engineering can significantly reduce it. We further propose a retrieval-augmented generation strategy to mitigate bias more effectively. Numerical experiments validate these strategies, demonstrating both the pervasive nature of bias and the impact of the proposed solutions.
Authors: Zhiteng Li, Xianglong Yan, Tianao Zhang, Haotong Qin, Dong Xie, Jiang Tian, zhongchao shi, Linghe Kong, Yulun Zhang, Xiaokang Yang
Abstract: Large Language Models (LLMs) have greatly pushed forward advancements in natural language processing, yet their high memory and computational demands hinder practical deployment. Binarization, as an effective compression technique, can shrink model weights to just 1 bit, significantly reducing the high demands on computation and memory. However, current binarization methods struggle to narrow the distribution gap between binarized and full-precision weights, while also overlooking the column deviation in LLM weight distribution. To tackle these issues, we propose ARB-LLM, a novel 1-bit post-training quantization (PTQ) technique tailored for LLMs. To narrow the distribution shift between binarized and full-precision weights, we first design an alternating refined binarization (ARB) algorithm to progressively update the binarization parameters, which significantly reduces the quantization error. Moreover, considering the pivot role of calibration data and the column deviation in LLM weights, we further extend ARB to ARB-X and ARB-RC. In addition, we refine the weight partition strategy with column-group bitmap (CGB), which further enhance performance. Equipping ARB-X and ARB-RC with CGB, we obtain ARB-LLM$_\text{X}$ and ARB-LLM$_\text{RC}$ respectively, which significantly outperform state-of-the-art (SOTA) binarization methods for LLMs. As a binary PTQ method, our ARB-LLM$_\text{RC}$ is the first to surpass FP16 models of the same size. The code and models will be available at https://github.com/ZHITENGLI/ARB-LLM.
Authors: Chen-Yang Xu, Han-Guang Wang, Lan Zhang, Yong-Hui Zhang, Hui-Rang Hou, Qing-Hao Meng
Abstract: Recently, researchers have begun to experiment with deep learning-based methods for detecting major depressive disor-der (MDD) using electroencephalogram (EEG) signals in search of a more objective means of diagnosis. However, exist-ing spatiotemporal feature extraction methods only consider the functional correlation between multiple electrodes and temporal correlation of EEG signals, ignoring the spatial posi-tion connection information between electrodes and the conti-nuity between time windows, which reduces the model's fea-ture extraction capabilities. To address this issue, a Spatio-temporal fused network for MDD detection with Electrode spatial Topology and adjacent TIME-window transition in-formation (SET-TIME) is proposed in this study. SET-TIME is composed by a common feature extractor, a secondary time-correlation feature extractor, and a domain adaptation (DA) module, in which the former extractor is used to obtain the temporal and spatial features, and the latter extractor can mine the correlation between multiple time windows, and the DA module is adopted to enhance cross-subject detection ca-pability. The experimental results of 10-fold cross-validation show that the proposed SET-TIME method outperforms the state-of-the-art (SOTA) method by achieving MDD detection accuracies of 92.00% and 94.00% on the public datasets PRED+CT and MODMA, respectively. Ablation experiments demonstrate the effectiveness of the multiple modules in SET-TIME, which assist in MDD detection by exploring the intrin-sic spatiotemporal information of EEG signals.
Authors: Jean Kossaifi, Nikola Kovachki, Zongyi Li, David Pitt, Miguel Liu-Schiaffini, Robert Joseph George, Boris Bonev, Kamyar Azizzadenesheli, Julius Berner, Valentin Duruisseaux, Anima Anandkumar
Abstract: We present NeuralOperator, an open-source Python library for operator learning. Neural operators generalize neural networks to maps between function spaces instead of finite-dimensional Euclidean spaces. They can be trained and inferenced on input and output functions given at various discretizations, satisfying a discretization convergence properties. Part of the official PyTorch Ecosystem, NeuralOperator provides all the tools for training and deploying neural operator models, as well as developing new ones, in a high-quality, tested, open-source package. It combines cutting-edge models and customizability with a gentle learning curve and simple user interface for newcomers.
Authors: Bo Gao, Michael W. Spratling, Letizia Gionfrida
Abstract: Large language models have achieved remarkable success in recent years, primarily due to self-attention. However, traditional Softmax attention suffers from numerical instability and reduced performance as the number of inference tokens increases. This work addresses these issues by proposing a new design principle for attention, viewing it as a two-stage process. The first stage (normalisation) refines standard attention by replacing Softmax with the more numerically stable Softplus followed by $l_{1}$-normalisation. Furthermore, we introduce a dynamic scale factor based on invariance entropy. We show that this novel attention mechanism outperforms conventional Softmax attention, and state-of-the-art Softmax-free alternatives. Our second proposal is to introduce a second processing stage (sharpening) which consists of a re-weighting mechanism that amplifies significant attentional weights while diminishing weaker ones. This enables the model to concentrate more effectively on relevant tokens, mitigating the attention sink phenomenon, and fundamentally improving length extrapolation. This novel, two-stage, replacement for self-attention is shown to ensure numerical stability and dramatically improve length extrapolation, maintaining a nearly constant validation loss at 16$\times$ the training length while achieving superior results on challenging long-context retrieval tasks and downstream benchmarks. Furthermore, symbolic regression experiments demonstrate that our method enables models to recover Newton's gravitational law from orbital trajectory sequences, providing evidence that appropriate attention mechanisms are crucial for foundation models to develop genuine physical world models.
Authors: Akiyoshi Tomihari, Issei Sato
Abstract: Transformers are difficult to optimize with stochastic gradient descent (SGD) and largely rely on adaptive optimizers such as Adam. Despite their empirical success, the reasons behind Adam's superior performance over SGD remain poorly understood. In this study, we analyze the optimization of Transformer models through the lens of \emph{gradient heterogeneity}, defined as the variation in gradient norms across parameter blocks. We provide a theoretical analysis showing that gradient heterogeneity, together with Hessian heterogeneity, degrades the convergence of gradient-based methods such as SGD, while sign-based methods are substantially less sensitive to this effect. Adam's coordinate-wise normalization makes its update directions depend mainly on gradient signs, so Adam can be interpreted as a soft variant of SignSGD. Our analysis uses the fact that SGD and SignSGD follow steepest descent directions under different norms, and derives upper bounds on the iteration complexity with implications for learning rate scaling in SignSGD. We further investigate the origin of gradient heterogeneity in Transformer architectures and show that it is strongly influenced by the placement of layer normalization, with Post-LN architectures exhibiting particularly pronounced heterogeneity. Experimental results from fine-tuning Transformers in both NLP and vision domains validate our theoretical analysis. Code is available at https://github.com/tom4649/gradient-heterogeneity.
Authors: Daqian Shao, Thomas Kleine Buening, Marta Kwiatkowska
Abstract: We propose a general framework for causal Imitation Learning (IL) with hidden confounders, which subsumes several existing settings. Our framework accounts for two types of hidden confounders: (a) variables observed by the expert but not by the imitator, and (b) confounding noise hidden from both. By leveraging trajectory histories as instruments, we reformulate causal IL in our framework into a Conditional Moment Restriction (CMR) problem. We propose DML-IL, an algorithm that solves this CMR problem via instrumental variable regression, and upper bound its imitation gap. Empirical evaluation on continuous state-action environments, including Mujoco tasks, demonstrates that DML-IL outperforms existing causal IL baselines.
Authors: Austin Shouli, Ankur Barthwal, Molly Campbell, Ajay Kumar Shrestha
Abstract: The rapid expansion of Artificial Intelligence (AI) in digital platforms used by youth has created significant challenges related to privacy, autonomy, and data protection. While AI-driven personalization offers enhanced user experiences, it often operates without clear ethical boundaries, leaving young users vulnerable to data exploitation and algorithmic biases. This paper presents a call to action for ethical AI governance, advocating for a structured framework that ensures youth-centred privacy protections, transparent data practices, and regulatory oversight. We outline key areas requiring urgent intervention, including algorithmic transparency, privacy education, parental data-sharing ethics, and accountability measures. Through this approach, we seek to empower youth with greater control over their digital identities and propose actionable strategies for policymakers, AI developers, and educators to build a fairer and more accountable AI ecosystem.
Authors: Albert Sawczyn, Jakub Binkowski, Denis Janiak, Bogdan Gabrys, Tomasz Kajdanowicz
Abstract: Large Language Models (LLMs) frequently generate hallucinated content, posing significant challenges for applications where factuality is crucial. While existing hallucination detection methods typically operate at the sentence level or passage level, we propose FactSelfCheck, a novel zero-resource black-box sampling-based method that enables fine-grained fact-level detection. Our approach represents text as interpretable knowledge graphs consisting of facts in the form of triples, providing clearer insights into content factuality than traditional approaches. Through analyzing factual consistency across multiple LLM responses, we compute fine-grained hallucination scores without requiring external resources or training data. Our evaluation demonstrates that FactSelfCheck performs competitively with leading sentence-level sampling-based methods while providing more detailed and interpretable insights. Most notably, our fact-level approach significantly improves hallucination correction, achieving a 35.5% increase in factual content compared to the baseline, while sentence-level SelfCheckGPT yields only a 10.6% improvement. The granular nature of our detection enables more precise identification and correction of hallucinated content. Additionally, we contribute FavaMultiSamples, a novel dataset that addresses a gap in the field by providing the research community with a second dataset for evaluating sampling-based methods.
Authors: Xiangwen Zhang, Qian Zhang, Longfei Han, Qiang Qu, Xiaoming Chen, Weidong Cai
Abstract: Collecting real-world vehicle accident videos for autonomous driving research is challenging due to their rarity and complexity. While existing driving video generation methods may produce visually realistic videos, they often fail to deliver physically realistic simulations because they lack the capability to generate accurate post-collision trajectories. In this paper, we introduce AccidentSim, a novel framework that generates physically realistic vehicle collision videos by extracting and utilizing the physical clues and contextual information available in real-world vehicle accident reports. Specifically, AccidentSim leverages a reliable physical simulator to replicate post-collision vehicle trajectories from the physical and contextual information in the accident reports and to build a vehicle collision trajectory dataset. This dataset is then used to fine-tune a language model, enabling it to respond to user prompts and predict physically consistent post-collision trajectories across various driving scenarios based on user descriptions. Finally, we employ Neural Radiance Fields (NeRF) to render high-quality backgrounds, merging them with the foreground vehicles that exhibit physically realistic trajectories to generate vehicle collision videos. Experimental results demonstrate that the videos produced by AccidentSim excel in both visual and physical authenticity.
Authors: Shahryar Zehtabi, Dong-Jun Han, Seyyedali Hosseinalipour, Christopher G. Brinton
Abstract: Much of federated learning (FL) focuses on settings where local dataset statistics remain the same between training and testing. However, this assumption often does not hold in practice due to distribution shifts, motivating the development of domain generalization (DG) approaches that leverage source domain data to train models capable of generalizing to unseen target domains. In this paper, we are motivated by two major gaps in existing work on FL and DG: (1) the lack of formal mathematical analysis of DG objectives; and (2) DG research in FL being limited to the star-topology architecture. We develop Decentralized Federated Domain Generalization with Style Sharing ($\textit{StyleDDG}$), a decentralized DG algorithm which allows devices in a peer-to-peer network to achieve DG based on sharing style information inferred from their datasets. Additionally, we provide the first systematic approach to analyzing style-based DG training in decentralized networks. We cast existing centralized DG algorithms within our framework, and employ their formalisms to model $\textit{StyleDDG}$. We then obtain analytical conditions under which convergence of $\textit{StyleDDG}$ can be guaranteed. Through experiments on popular DG datasets, we demonstrate that $\textit{StyleDDG}$ can obtain significant improvements in accuracy across target domains with minimal communication overhead compared to baseline decentralized gradient methods.
Authors: Patrick Haller, Jonas Golde, Alan Akbik
Abstract: Linearization has emerged as a strategy for developing efficient language models (LMs). Starting from an existing Transformer-based LM, linearization replaces the attention component with computationally efficient subquadratic \textit{token mixers}. However, as an increasing number of mixers are proposed, it remains unclear which inductive biases are best suited to inherit the original Transformer's capabilities. Furthermore, it is unknown how linearization is affected by parameter and token budget scaling. To address these questions, we propose a unified setup to compare seven representative architectures, including xLSTM, GLA, and Gated DeltaNet. Our findings reveal that performance hierarchies remain stable from 140M to 1.7B parameters, with error-correcting update rules demonstrating superior scaling exponents. We show that performance gaps are established early and persist through asymptotic maturity at 10B tokens, suggesting that state resolution is a more fundamental bottleneck than the distillation budget. Finally, while most models adapt to instruction tuning, only gated delta-rule formulations maintain the precision necessary for long-context retrieval, whereas additive models suffer from irreversible state saturation. These results suggest that for successful linearization, architectural inductive biases remain the primary constraint that cannot be overcome by simply scaling training compute.
Authors: Thomas Nagler, David R\"ugamer
Abstract: Prior-data fitted networks (PFNs) have emerged as promising foundation models for prediction from tabular data sets, achieving state-of-the-art performance on small to moderate data sizes without tuning. While PFNs are motivated by Bayesian ideas, they do not provide any uncertainty quantification for predictive means, quantiles, or similar quantities. We propose a principled and efficient sampling procedure to construct Bayesian posteriors for such estimates based on Martingale posteriors, and prove its convergence. Several simulated and real-world data examples showcase the uncertainty quantification of our method in inference applications.
Authors: Anjiang Wei, Tarun Suresh, Huanmi Tan, Yinglun Xu, Gagandeep Singh, Ke Wang, Alex Aiken
Abstract: Superoptimization is the task of transforming a program into a faster one while preserving its input-output behavior. In this work, we investigate whether large language models (LLMs) can serve as superoptimizers, generating assembly programs that outperform code already optimized by industry-standard compilers. We construct the first large-scale benchmark for this problem, consisting of 8,072 assembly programs averaging 130 lines, in contrast to prior datasets restricted to 2-15 straight-line, loop-free programs. We evaluate 23 LLMs on this benchmark and find that the strongest baseline, Claude-opus-4, achieves a 51.5% test-passing rate and a 1.43x average speedup over gcc -O3. To further enhance performance, we fine-tune models with reinforcement learning, optimizing a reward function that integrates correctness and performance speedup. Starting from Qwen2.5-Coder-7B-Instruct (61.4% correctness, 1.10x speedup), the fine-tuned model SuperCoder attains 95.0% correctness and 1.46x average speedup, with additional improvement enabled by Best-of-N sampling and iterative refinement. Our results demonstrate, for the first time, that LLMs can be applied as superoptimizers for assembly programs, establishing a foundation for future research in program performance optimization beyond compiler heuristics.
Authors: Matthew Landers, Taylor W. Killian, Thomas Hartvigsen, Afsaneh Doryab
Abstract: The combinatorial structure of many real-world action spaces leads to exponential growth in the number of possible actions, limiting the effectiveness of conventional reinforcement learning algorithms. Recent approaches for combinatorial action spaces impose factorized or sequential structures over sub-actions, failing to capture complex joint behavior. We introduce the Sub-Action Interaction Network using Transformers (SAINT), a novel policy architecture that represents multi-component actions as unordered sets and models their dependencies via self-attention conditioned on the global state. SAINT is permutation-invariant, sample-efficient, and compatible with standard policy optimization algorithms. In 18 distinct combinatorial environments across three task domains, including environments with $1.35 \times 10^{18}$ possible actions, SAINT consistently outperforms strong baselines.
Authors: Guangyuan Ma, Yongliang Ma, Xuanrui Gou, Zhenpeng Su, Ming Zhou, Songlin Hu
Abstract: Large Language Models (LLMs)-based text retrieval retrieves documents relevant to search queries based on vector similarities. Documents are pre-encoded offline, while queries arrive in real-time, necessitating an efficient online query encoder. Although LLMs significantly enhance retrieval capabilities, serving deeply parameterized LLMs slows down query inference throughput and increases demands for online deployment resources. In this paper, we propose LightRetriever, a novel LLM-based retriever with extremely lightweight query encoders. Our method retains a full-sized LLM for document encoding, but reduces the workload of query encoding to no more than an embedding lookup. Compared to serving a full LLM on an A800 GPU, our method achieves over 1000x speedup in query encoding and over 10x increase in end-to-end retrieval throughput. Extensive experiments on large-scale retrieval benchmarks show that LightRetriever generalizes well across diverse tasks, maintaining an average of 95% retrieval performance.
Authors: Jiayu Chen, Le Xu, Aravind Venugopal, Jeff Schneider
Abstract: Offline reinforcement learning (RL) offers a powerful paradigm for data-driven control. Compared to model-free approaches, offline model-based RL (MBRL) explicitly learns a world model from a static dataset and uses it as a surrogate simulator, improving data efficiency and enabling potential generalization beyond the dataset support. However, most existing offline MBRL methods follow a two-stage training procedure: first learning a world model by maximizing the likelihood of the observed transitions, then optimizing a policy to maximize its expected return under the learned model. This objective mismatch results in a world model that is not necessarily optimized for effective policy learning. Moreover, we observe that policies learned via offline MBRL often lack robustness during deployment, and small adversarial noise in the environment can lead to significant performance degradation. To address these, we propose a framework that dynamically adapts the world model alongside the policy under a unified learning objective aimed at improving robustness. At the core of our method is a maximin optimization problem, which we solve by innovatively utilizing Stackelberg learning dynamics. We provide theoretical analysis to support our design and introduce computationally efficient implementations. We benchmark our algorithm on twelve noisy D4RL MuJoCo tasks and three stochastic Tokamak Control tasks, demonstrating its state-of-the-art performance.
Authors: Aryaman Arora, Neil Rathi, Nikil Roashan Selvam, R\'obert Csord\'as, Dan Jurafsky, Christopher Potts
Abstract: State space models (SSMs) for language modelling promise an efficient and performant alternative to quadratic-attention Transformers, yet show variable performance on recalling basic information from the context. While performance on synthetic tasks like Associative Recall (AR) can point to this deficiency, behavioural metrics provide little information as to \textit{why} -- on a mechanistic level -- certain architectures fail and others succeed. To address this, we conduct experiments on AR, and find that only Transformers and Based SSM models fully succeed at AR, with Mamba and DeltaNet close behind, while the other SSMs (H3, Hyena) fail. We then use causal interventions to explain why. We find that Transformers and Based learn to store key-value associations in-context using induction. By contrast, the SSMs seem to compute these associations only at the last state using a single layer. We further investigate the mechanism underlying the success of Mamba, and find novel evidence that Mamba \textit{does} implement induction: not via the SSM, but instead via short convolutions. Further experiments on a new hierarchical retrieval task, Associative Treecall (ATR), show that all architectures learn the same mechanism as they did for AR. Furthermore, we show that Mamba can learn Attention-like induction on ATR when short convolutions are removed. These results reveal that architectures with similar accuracy may still have substantive differences, motivating the adoption of mechanistic evaluations.
Authors: Seonghwan Park, Jueun Mun, Donghyun Oh, Namhoon Lee
Abstract: Concept bottleneck models (CBMs) ensure interpretability by decomposing predictions into human interpretable concepts. Yet the annotations used for training CBMs that enable this transparency are often noisy, and the impact of such corruption is not well understood. In this study, we present the first systematic study of noise in CBMs and show that even moderate corruption simultaneously impairs prediction performance, interpretability, and the intervention effectiveness. Our analysis identifies a susceptible subset of concepts whose accuracy declines far more than the average gap between noisy and clean supervision and whose corruption accounts for most performance loss. To mitigate this vulnerability we propose a two-stage framework. During training, sharpness-aware minimization stabilizes the learning of noise-sensitive concepts. During inference, where clean labels are unavailable, we rank concepts by predictive entropy and correct only the most uncertain ones, using uncertainty as a proxy for susceptibility. Theoretical analysis and extensive ablations elucidate why sharpness-aware training confers robustness and why uncertainty reliably identifies susceptible concepts, providing a principled basis that preserves both interpretability and resilience in the presence of noise.
Authors: Deyang Kong, Qi Guo, Xiangyu Xi, Wei Wang, Jingang Wang, Xunliang Cai, Shikun Zhang, Wei Ye
Abstract: Reinforcement learning exhibits potential in enhancing the reasoning abilities of large language models, yet it is hard to scale for the low sample efficiency during the rollout phase. Existing methods attempt to improve efficiency by scheduling problems based on problem difficulties. However, these approaches suffer from unstable and biased estimations of problem difficulty and fail to capture the alignment between model competence and problem difficulty in RL training, leading to suboptimal results. To tackle these limitations, this paper introduces $\textbf{C}$ompetence-$\textbf{D}$ifficulty $\textbf{A}$lignment $\textbf{S}$ampling ($\textbf{CDAS}$), which enables accurate and stable estimation of problem difficulties by aggregating historical performance discrepancies of problems. Then the model competence is quantified to adaptively select problems whose difficulty is in alignment with the model's current competence using a fixed-point system. Experimental results across a range of challenging mathematical benchmarks show that CDAS achieves great improvements in both accuracy and efficiency. CDAS attains the highest average accuracy against baselines and exhibits significant speed advantages compared to Dynamic Sampling, a competitive strategy in DAPO, which is 2.33 times slower than CDAS.
Authors: Achkan Salehi
Abstract: Ballbot (i.e. Ball balancing robot) navigation usually relies on methods rooted in control theory (CT), and works that apply Reinforcement learning (RL) to the problem remain rare while generally being limited to specific subtasks (e.g. balance recovery). Unlike CT based methods, RL does not require (simplifying) assumptions about environment dynamics (e.g. the absence of slippage between the ball and the floor). In addition to this increased accuracy in modeling, RL agents can easily be conditioned on additional observations such as depth-maps without the need for explicit formulations from first principles, leading to increased adaptivity. Despite those advantages, there has been little to no investigation into the capabilities, data-efficiency and limitations of RL based methods for ballbot control and navigation. Furthermore, there is a notable absence of an open-source, RL-friendly simulator for this task. In this paper, we present an open-source ballbot simulation based on MuJoCo, and show that with appropriate conditioning on exteroceptive observations as well as reward shaping, policies learned by classical model-free RL methods are capable of effectively navigating through randomly generated uneven terrain, using a reasonable amount of data (four to five hours on a system operating at 500hz). Our code is made publicly available.
Authors: Afrozah Nadeem, Mark Dras, Usman Naseem
Abstract: Large Language Models (LLMs) increasingly shape public discourse, yet most evaluations of political and economic bias have focused on high-resource, Western languages and contexts. This leaves critical blind spots in low-resource, multilingual regions such as Pakistan, where linguistic identity is closely tied to political, religious, and regional ideologies. We present a systematic evaluation of political bias in 13 state-of-the-art LLMs across five Pakistani languages: Urdu, Punjabi, Sindhi, Pashto, and Balochi. Our framework integrates a culturally adapted Political Compass Test (PCT) with multi-level framing analysis, capturing both ideological stance (economic/social axes) and stylistic framing (content, tone, emphasis). Prompts are aligned with 11 socio-political themes specific to the Pakistani context. Results show that while LLMs predominantly reflect liberal-left orientations consistent with Western training data, they exhibit more authoritarian framing in regional languages, highlighting language-conditioned ideological modulation. We also identify consistent model-specific bias patterns across languages. These findings show the need for culturally grounded, multilingual bias auditing frameworks in global NLP.
Authors: SeungBum Ha, Saerom Park, Sung Whan Yoon
Abstract: Machine unlearning (MU) aims to expunge a designated forget set from a trained model without costly retraining, yet the existing techniques overlook two critical blind spots: "over-unlearning" that deteriorates retained data near the forget set, and post-hoc "relearning" attacks that aim to resurrect the forgotten knowledge. Focusing on class-level unlearning, we first derive an over-unlearning metric, OU@epsilon, which quantifies collateral damage in regions proximal to the forget set, where over-unlearning mainly appears. Next, we expose an unforeseen relearning threat on MU, i.e., the Prototypical Relearning Attack, which exploits the per-class prototype of the forget class with just a few samples, and easily restores the pre-unlearning performance. To counter both blind spots in class-level unlearning, we introduce Spotter, a plug-and-play objective that combines (i) a masked knowledge-distillation penalty on the nearby region of forget classes to suppress OU@epsilon, and (ii) an intra-class dispersion loss that scatters forget-class embeddings, neutralizing Prototypical Relearning Attacks. Spotter achieves state-of-the-art results across CIFAR, TinyImageNet, and CASIA-WebFace datasets, offering a practical remedy to unlearning's blind spots.
Authors: Jaeseung Heo, Kyeongheung Yun, Seokwon Yoon, MoonJeong Park, Jungseul Ok, Dongwoo Kim
Abstract: Understanding how individual edges influence the behavior of graph neural networks (GNNs) is essential for improving their interpretability and robustness. Graph influence functions have emerged as promising tools to efficiently estimate the effects of edge deletions without retraining. However, existing influence prediction methods rely on strict convexity assumptions, exclusively consider the influence of edge deletions while disregarding edge insertions, and fail to capture changes in message propagation caused by these modifications. In this work, we propose a proximal Bregman response function specifically tailored for GNNs, relaxing the convexity requirement and enabling accurate influence prediction for standard neural network architectures. Furthermore, our method explicitly accounts for message propagation effects and extends influence prediction to both edge deletions and insertions in a principled way. Experiments with real-world datasets demonstrate accurate influence predictions for different characteristics of GNNs. We further demonstrate that the influence function is versatile in applications such as graph rewiring and adversarial attacks.
Authors: Arian Raje, Baris Askin, Divyansh Jhunjhunwala, Gauri Joshi
Abstract: Large language models (LLMs) have not yet effectively leveraged the vast amounts of edge-device data, and federated learning (FL) offers a promising paradigm to collaboratively fine-tune LLMs without transferring private edge data to the cloud. To operate within the computation and communication constraints of edge devices, recent literature on federated fine-tuning of LLMs proposes the use of low-rank adaptation (LoRA) and similar parameter-efficient methods. However, LoRA-based methods suffer from accuracy degradation in FL settings, primarily because of data and computational heterogeneity across clients. We propose Ravan, an adaptive multi-head LoRA method that balances parameter efficiency and model expressivity by reparameterizing the weight updates as the sum of multiple LoRA heads $s_i\textbf{B}_i\textbf{H}_i\textbf{A}_i$ in which only the core matrices $\textbf{H}_i$ and their lightweight scaling factors $s_i$ are trained. These trainable scaling factors let the optimization focus on the most useful heads, recovering a higher-rank approximation of the full update without increasing the number of communicated parameters since clients upload $s_i\textbf{H}_i$ directly. Experiments on vision and language benchmarks show that Ravan improves test accuracy by $2-8\%$ over prior parameter-efficient baselines, making it a robust and scalable solution for federated fine-tuning of LLMs.
Authors: Ivi Chatzi, Nina Corvelo Benz, Stratis Tsirtsis, Manuel Gomez-Rodriguez
Abstract: Providers of LLM-as-a-service have predominantly adopted a simple pricing model: users pay a fixed price per token. Consequently, one may think that the price two different users would pay for the same output string under the same input prompt is the same. In our work, we show that, surprisingly, this is not (always) true. We find empirical evidence that, particularly for non-english outputs, both proprietary and open-weights LLMs often generate the same (output) string with multiple different tokenizations, even under the same input prompt, and this in turn leads to arbitrary price variation. To address the problem of tokenization multiplicity, we introduce canonical generation, a type of constrained generation that restricts LLMs to only generate canonical tokenizations -- the unique tokenization in which each string is tokenized during the training process of an LLM. Further, we introduce an efficient sampling algorithm for canonical generation based on the Gumbel-Max trick. Experiments on a variety of natural language tasks demonstrate that our sampling algorithm for canonical generation is comparable to standard sampling in terms of performance and runtime, and it solves the problem of tokenization multiplicity.
Authors: Francisco Caetano, Christiaan Viviers, Peter H. N. De With, Fons van der Sommen
Abstract: Flow Matching has emerged as a powerful framework for learning continuous transformations between distributions, enabling high-fidelity generative modeling. This work introduces Symmetrical Flow Matching (SymmFlow), a new formulation that unifies semantic segmentation, classification, and image generation within a single model. Using a symmetric learning objective, SymmFlow models forward and reverse transformations jointly, ensuring bi-directional consistency, while preserving sufficient entropy for generative diversity. A new training objective is introduced to explicitly retain semantic information across flows, featuring efficient sampling while preserving semantic structure, allowing for one-step segmentation and classification without iterative refinement. Unlike previous approaches that impose strict one-to-one mapping between masks and images, SymmFlow generalizes to flexible conditioning, supporting both pixel-level and image-level class labels. Experimental results on various benchmarks demonstrate that SymmFlow achieves state-of-the-art performance on semantic image synthesis, obtaining FID scores of 11.9 on CelebAMask-HQ and 7.0 on COCO-Stuff with only 25 inference steps. Additionally, it delivers competitive results on semantic segmentation and shows promising capabilities in classification tasks.
Authors: Chi Zuo, Martin B. M{\o}ller, Pablo Mart\'inez-Nuevo, Huayang Huang, Yu Wu, Ye Zhu
Abstract: While being disturbed by environmental noises, the acoustic masking technique is a conventional way to reduce the annoyance in audio engineering that seeks to cover up the noises with other dominant yet less intrusive sounds. However, misalignment between the dominant sound and the noise-such as mismatched downbeats-often requires an excessive volume increase to achieve effective masking. Motivated by recent advances in cross-modal generation, in this work, we introduce an alternative method to acoustic masking, aiming to reduce the noticeability of environmental noises by blending them into personalized music generated based on user-provided text prompts. Following the paradigm of music generation using mel-spectrogram representations, we propose a Blending Noises into Personalized Music (BNMusic) framework with two key stages. The first stage synthesizes a complete piece of music in a mel-spectrogram representation that encapsulates the musical essence of the noise. In the second stage, we adaptively amplify the generated music segment to further reduce noise perception and enhance the blending effectiveness, while preserving auditory quality. Our experiments with comprehensive evaluations on MusicBench, EPIC-SOUNDS, and ESC-50 demonstrate the effectiveness of our framework, highlighting the ability to blend environmental noise with rhythmically aligned, adaptively amplified, and enjoyable music segments, minimizing the noticeability of the noise, thereby improving overall acoustic experiences. Project page: https://d-fas.github.io/BNMusic_page/.
Authors: Yifei Xu, Tusher Chakraborty, Srinagesh Sharma, Leonardo Nunes, Swati Sharma, Kate Drakos Demopulos, Emre K{\i}c{\i}man, Songwu Lu, Ranveer Chandra
Abstract: RL training of LLMs on open-ended tasks is challenging due to the lack of direct verifiability. In this paper, we frame such training as constrained RL that (i) optimizes a token-level dense Reasoning Reflection Reward (R3) aligned with reasoning quality, and (ii) enforces rubric-gating as feasibility constraints at the rollout group level. R3 measures the model's token-level certainty of a reference answer under its CoT reasoning prefix while selectively emphasizing reasoning-reflective tokens to capture how likely the generated reasoning is to yield the desired answer. Rubric-gating complements R3 by operationalizing principled task criteria as hard accept/reject checks on final answers. Empirically, across four datasets, our framework outperforms baselines, achieves faster, more sample-efficient learning, and respects feasibility constraints.
Authors: Wei Shi, Ziyuan Xie, Sihang Li, Xiang Wang
Abstract: Reinforcement learning from human feedback (RLHF) is a key paradigm for aligning large language models (LLMs) with human values, yet the reward models at its core remain largely opaque. In this work, we present Sparse Autoencoder For Enhanced Reward model (\textbf{SAFER}), a novel framework for interpreting and improving reward models through mechanistic analysis. Leveraging Sparse Autoencoders (SAEs), we uncover human-interpretable features in reward model activations, enabling insight into safety-relevant decision-making. We apply SAFER to safety-oriented preference datasets and quantify the salience of individual features by activation differences between chosen and rejected responses. Using these feature-level signals, we design targeted data poisoning and denoising strategies. Experiments show that SAFER can precisely degrade or enhance safety alignment with minimal data modification, without sacrificing general chat performance. Our approach contributes to interpreting, auditing and refining reward models in high-stakes LLM alignment tasks. Our codes are available at https://github.com/xzy-101/SAFER-code. \textit{This paper discusses topics related to reward model safety and may include discussions or examples that highlight potential risks or unsafe outcomes.}
Authors: Bo Yan, Yurong Hao, Dingqi Liu, Huabin Sun, Pengpeng Qiao, Wei Yang Bryan Lim, Yang Cao, Chuan Shi
Abstract: Federated recommender systems (FedRec) have emerged as a promising approach to provide personalized recommendations while protecting user privacy. However, recent studies have shown their vulnerability to poisoning attacks, where malicious clients inject crafted gradients to promote target items to benign users. Existing attacks typically target the full user group, which compromises stealth and increases detection risk. In contrast, real-world adversaries may prefer to target specific user subgroups, such as promoting health supplements to older individuals, to maximize effectiveness while preserving stealth. Motivated by this gap, we introduce Spattack, the first poisoning attack designed to manipulate recommendations for specific user subgroups in federated settings. Spattack adopts an approximate-and-promote paradigm, which approximates user embeddings of target and non-target subgroups and then promotes target items to the target subgroup. We further reveal a trade-off between strong attack performance on the target subgroup and limited impact on the non-target subgroup. To achieve a better trade-off, we propose enhanced approximation and promotion strategies. For approximation, we push embeddings of different subgroups apart via contrastive learning and augment the target subgroup's relevant item set through clustering. For promotion, we align embeddings of target items and relevant items to strengthen their semantic connections, together with an adaptive weighting strategy to balance effects across subgroups. Experiments on three real-world datasets demonstrate that Spattack achieves strong attack performance on the target subgroup with minimal impact on non-target users, even when only 0.1% of users are malicious. Moreover, Spattack maintains competitive recommendation performance and shows strong resilience against mainstream defenses.
Authors: Quang Truong, Zhikai Chen, Mingxuan Ju, Tong Zhao, Neil Shah, Jiliang Tang
Abstract: Relational databases underpin critical infrastructure across a wide range of domains, yet the design of generalizable pre-training strategies for learning from relational databases remains an open challenge due to task heterogeneity. Specifically, there exist many possible downstream tasks, as tasks are defined based on relational schema graphs, temporal dependencies, and SQL-defined label logics. An effective pre-training framework is desired to take these factors into account in order to obtain task-aware representations. By incorporating knowledge of the underlying distribution that drives label generation, downstream tasks can benefit from relevant side-channel information. To bridge this gap, we introduce Task Vector Estimation (TVE), a novel pre-training framework that constructs predictive supervisory signals via set-based aggregation over schema traversal graphs, explicitly modeling next-window relational dynamics. We formalize our approach through an information-theoretic lens, demonstrating that task-informed representations retain more relevant signals than those obtained without task priors. Extensive experiments on the RelBench benchmark show that TVE consistently outperforms traditional pre-training baselines. Our findings advocate for pre-training objectives that encode task heterogeneity and temporal structure as design principles for predictive modeling on relational databases. Our code is publicly available at https://github.com/quang-truong/task-vector-estimation.
URLs: https://github.com/quang-truong/task-vector-estimation.
Authors: Han Jiang, Dongyao Zhu, Zhihua Wei, Xiaoyuan Yi, Ziang Xiao, Xing Xie
Abstract: In-Context Learning has shown great potential for aligning Large Language Models (LLMs) with human values, helping reduce harmful outputs and accommodate diverse preferences without costly post-training, known as In-Context Alignment (ICA). However, LLMs' comprehension of input prompts remains agnostic, limiting ICA's ability to address value tensions--human values are inherently pluralistic, often imposing conflicting demands, e.g., stimulation vs. tradition. Current ICA methods therefore face the Instruction Bottleneck challenge, where LLMs struggle to reconcile multiple intended values within a single prompt, leading to incomplete or biased alignment. To address this, we propose PICACO, a novel pluralistic ICA method. Without fine-tuning, PICACO optimizes a meta-instruction that navigates multiple values to better elicit LLMs' understanding of them and improve their alignment. This is achieved by maximizing the total correlation between specified values and LLM responses, theoretically reinforcing value correlation while reducing distractive noise, resulting in effective value instructions. Extensive experiments on five value sets show that PICACO works well with both black-box and open-source LLMs, outperforms several recent strong baselines, and achieves a better balance across up to 8 distinct values.
Authors: Jinzhi Wang, Qingke Peng, Haozhou Li, Zeyuan Zeng, Jiangbo Zhang, Kaixuan Yang, Ningyong Wu, Qinfeng Song, Ruimeng Li, Biyi Zhou
Abstract: As power systems decarbonise and digitalise, high penetrations of distributed energy resources and flexible tariffs make electric power marketing (EPM) a key interface between regulation, system operation and sustainable-energy deployment. Many utilities still rely on human agents and rule- or intent-based chatbots with fragmented knowledge bases that struggle with long, cross-scenario dialogues and fall short of requirements for compliant, verifiable and DR-ready interactions. Meanwhile, frontier large language models (LLMs) show strong conversational ability but are evaluated on generic benchmarks that underweight sector-specific terminology, regulatory reasoning and multi-turn process stability. To address this gap, we present ElectriQ, a large-scale benchmark and evaluation framework for LLMs in EPM. ElectriQ contains over 550k dialogues across six service domains and 24 sub-scenarios and defines a unified protocol that combines human ratings, automatic metrics and two compliance stress tests-Statutory Citation Correctness and Long-Dialogue Consistency. Building on ElectriQ, we propose SEEK-RAG, a retrieval-augmented method that injects policy and domain knowledge during finetuning and inference. Experiments on 13 LLMs show that domain-aligned 7B models with SEEK-RAG match or surpass much larger models while reducing computational cost, providing an auditable, regulation-aware basis for deploying LLM-based EPM assistants that support demand-side management, renewable integration and resilient grid operation.
Authors: Sekh Mainul Islam, Nadav Borenstein, Siddhesh Milind Pawar, Haeun Yu, Arnav Arora, Isabelle Augenstein
Abstract: Understanding biases and stereotypes encoded in the weights of Large Language Models (LLMs) is crucial for developing effective mitigation strategies. However, biased behaviour is often subtle and non-trivial to isolate, even when deliberately elicited, making systematic analysis and debiasing particularly challenging. To address this, we introduce \texttt{BiasGym}, a simple, cost-effective, and generalizable framework for reliably and safely injecting, analyzing, and mitigating conceptual associations of biases within LLMs. \texttt{BiasGym} consists of two components: \texttt{BiasInject}, which safely injects specific biases into the model via token-based fine-tuning while keeping the model frozen, and \texttt{BiasScope}, which leverages these injected signals to identify and reliably steer the components responsible for biased behavior. Our method enables consistent bias elicitation for mechanistic analysis, supports targeted debiasing without degrading performance on downstream tasks, and generalizes to biases unseen during fine-tuning. We demonstrate the effectiveness of BiasGym in reducing real-world stereotypes (e.g., people from Italy being `reckless drivers'), showing its utility for both safety interventions and interpretability research.
Authors: Bing Han, Feifei Zhao, Dongcheng Zhao, Guobin Shen, Ping Wu, Yu Shi, Yi Zeng
Abstract: While fine-tuning services drive the rapid expansion of task capabilities in large language models (LLMs), they are often accompanied by the degradation and reorganization of safety-aligned representations, making models more prone to deviating from human preferences and exposing them to emerging jailbreak risks. Existing post-fine-tuning defense methods predominantly rely on single-scale safety correction mechanisms, which struggle to achieve a robust balance among safety, model utility, and continual adaptability. We propose Multi-Level Safety Continual Projection (MSCP), a training-free post-fine-tuning safety enhancement method that implicitly aligns global and localized safety activations through coordinated multi-level representations to isolate sparse neuron clusters governing safety-sensitive behaviors. It then applies composable safety-direction projections without retraining, effectively suppressing harmful outputs under minimal parameter perturbations while preserving task performance and improving alignment with human preferences. Extensive experiments across multiple fine-tuned LLM models demonstrate that our method significantly reduce harmfulness scores and attack success rates with minimal parameter modifications, while preserving the model's utility. Furthermore, we introduce a task-specific, multi-dimensional heterogeneous safety activation clustering mechanism that enables continual defense and generalization capability against unforeseen emerging safety concerns.
Authors: Yiqun Yao, Xiang Li, Xin Jiang, Xuezhi Fang, Naitong Yu, Wenjia Ma, Aixin Sun, Yequan Wang
Abstract: Full-duplex dialog models aim to listen and speak simultaneously, delivering rapid responses to dynamic user input. Among different solutions to full-duplexity, a native solution merges multiple channels in each time step, achieving the lowest latency. However, prevailing designs break down the textual monologue sentences for word-level alignment with audio streams, which degrades language modeling abilities. To help address this issue, we introduce "contiguous monologues", which are composed by continuous sentences and "waiting" intervals, mimicking human-like cognitive behavior in dialogs. We find a proper training paradigm to be critical for semantically aligning contiguous monologues with audio. To this end, we develop a "dual" training paradigm that alternates the position of the monologues, either leading or trailing the audio, across different training stages. A combination of our contiguous monologue and dual training strategy is applied in developing FLM-Audio, our 7B spoken dialog chatbot with native full-duplexity. As confirmed by experimental results, FLM-Audio achieves superior response qualities and chatting experiences while requiring significantly less training data.
Authors: Dominik Pegler, David Steyrl, Mengfan Zhang, Alexander Karner, Jozsef Arato, Frank Scharnowski, Filip Melinscak
Abstract: Phobias are common and impairing, and exposure therapy, which involves confronting patients with fear-provoking visual stimuli, is the most effective treatment. Scalable computerized exposure therapy requires automated prediction of fear directly from image content to adapt stimulus selection and treatment intensity. Whether such predictions can be made reliably and generalize across individuals and stimuli, however, remains unknown. Here we show that pretrained convolutional and transformer vision models, adapted via transfer learning, accurately predict group-level perceived fear for spider-related images, even when evaluated on new people and new images, achieving a mean absolute error (MAE) below 10 units on the 0-100 fear scale. Visual explanation analyses indicate that predictions are driven by spider-specific regions in the images. Learning-curve analyses show that transformer models are data efficient and approach performance saturation with the available data (~300 images). Prediction errors increase for very low and very high fear levels and within specific categories of images. These results establish transparent, data-driven fear estimation from images, laying the groundwork for adaptive digital mental health tools.
Authors: Einari Vaaras, Manu Airaksinen
Abstract: Feature space topology refers to the organization of samples within the feature space. Modifying this topology can be beneficial in machine learning applications, including dimensionality reduction, generative modeling, transfer learning, and robustness to adversarial attacks. This paper introduces a novel loss function, Hopkins loss, which leverages the Hopkins statistic to enforce a desired feature space topology, which is in contrast to existing topology-related methods that aim to preserve input feature topology. We evaluate the effectiveness of Hopkins loss on speech, text, and image data in two scenarios: classification and dimensionality reduction using nonlinear bottleneck autoencoders. Our experiments show that integrating Hopkins loss into classification or dimensionality reduction has only a small impact on classification performance while providing the benefit of modifying feature topology.
Authors: Daniyal Kabir Dar, Qiben Yan, Li Xiao, Arun Ross
Abstract: Adversarial perturbations in speech pose a serious threat to automatic speech recognition (ASR) and speaker verification by introducing subtle waveform modifications that remain imperceptible to humans but can significantly alter system outputs. While targeted attacks on end-to-end ASR models have been widely studied, the phonetic basis of these perturbations and their effect on speaker identity remain underexplored. In this work, we analyze adversarial audio at the phonetic level and show that perturbations exploit systematic confusions such as vowel centralization and consonant substitutions. These distortions not only mislead transcription but also degrade phonetic cues critical for speaker verification, leading to identity drift. Using DeepSpeech as our ASR target, we generate targeted adversarial examples and evaluate their impact on speaker embeddings across genuine and impostor samples. Results across 16 phonetically diverse target phrases demonstrate that adversarial audio induces both transcription errors and identity drift, highlighting the need for phonetic-aware defenses to ensure the robustness of ASR and speaker recognition systems.
Authors: Aniello Panariello, Daniel Marczak, Simone Magistri, Angelo Porrello, Bart{\l}omiej Twardowski, Andrew D. Bagdanov, Simone Calderara, Joost van de Weijer
Abstract: In this paper, we address the challenges associated with merging low-rank adaptations of large neural networks. With the rise of parameter-efficient adaptation techniques, such as Low-Rank Adaptation (LoRA), model fine-tuning has become more accessible. While fine-tuning models with LoRA is highly efficient, existing merging methods often sacrifice this efficiency by merging fully-sized weight matrices. We propose the Core Space merging framework, which enables the merging of LoRA-adapted models within a common alignment basis, thereby preserving the efficiency of low-rank adaptation while substantially improving accuracy across tasks. We further provide a formal proof that projection into Core Space ensures no loss of information and provide a complexity analysis showing the efficiency gains. Extensive empirical results demonstrate that Core Space significantly improves existing merging techniques and achieves state-of-the-art results on both vision and language tasks while utilizing a fraction of the computational resources. Codebase is available at https://github.com/apanariello4/core-space-merging.
Authors: Chenhui Hu, Pengfei Cao, Yubo Chen, Kang Liu, Jun Zhao
Abstract: The fundamental representational units (FRUs) of large language models (LLMs) remain undefined, limiting further understanding of their underlying mechanisms. In this paper, we introduce Atom Theory to systematically define, evaluate, and identify such FRUs, which we term atoms. Building on the atomic inner product (AIP), a non-Euclidean metric that captures the underlying geometry of LLM representations, we formally define atoms and propose two key criteria for ideal atoms: faithfulness ($R^2$) and stability ($q^*$). We further prove that atoms are identifiable under threshold-activated sparse autoencoders (TSAEs). Empirically, we uncover a pervasive representation shift in LLMs and demonstrate that the AIP corrects this shift to capture the underlying representational geometry, thereby grounding Atom Theory. We find that two widely used units, neurons and features, fail to qualify as ideal atoms: neurons are faithful ($R^2=1$) but unstable ($q^*=0.5\%$), while features are more stable ($q^*=68.2\%$) but unfaithful ($R^2=48.8\%$). To find atoms of LLMs, leveraging atom identifiability under TSAEs, we show via large-scale experiments that reliable atom identification occurs only when the TSAE capacity matches the data scale. Guided by this insight, we identify FRUs with near-perfect faithfulness ($R^2=99.9\%$) and stability ($q^*=99.8\%$) across layers of Gemma2-2B, Gemma2-9B, and Llama3.1-8B, satisfying the criteria of ideal atoms statistically. Further analysis confirms that these atoms align with theoretical expectations and exhibit substantially higher monosemanticity. Overall, we propose and validate Atom Theory as a foundation for understanding the internal representations of LLMs. Code available at https://github.com/ChenhuiHu/towards_atoms.
Authors: Ivan Stresec, Joana P. Gon\c{c}alves
Abstract: Unsupervised black-box models are drivers of scientific discovery, yet are difficult to interpret, as their output is often a multidimensional embedding rather than a well-defined target. While explainability for supervised learning uncovers how input features contribute to predictions, its unsupervised counterpart should relate input features to the structure of the learned embeddings. However, adaptations of supervised model explainability for unsupervised learning provide either single-sample or dataset-summary explanations, remaining too fine-grained or reductive to be meaningful, and cannot explain embeddings without mapping functions. To bridge this gap, we propose LAVA, a post-hoc model-agnostic method to explain local embedding organization through feature covariation in the original input data. LAVA explanations comprise modules, capturing local subpatterns of input feature correlation that reoccur globally across the embeddings. LAVA delivers stable explanations at a desired level of granularity, revealing domain-relevant patterns such as visual parts of images or disease signals in cellular processes, otherwise missed by existing methods.
Authors: Madeleine Dwyer, Adam Sobey, Adriane Chapman
Abstract: Training large language models (LLMs) with reinforcement learning (RL) methods such as PPO and GRPO commonly relies on ratio clipping to stabilise updates. While effective at preventing instability, clipping discards information, introduces gradient discontinuities and can prevent exploration of better policies. Inspired by label smoothing, we propose Probability Smoothing Policy Optimisation (PSPO). PSPO smooths current policy probabilities toward the behaviour policy before computing importance ratios, creating a soft trust region that preserves gradients while preventing destabilising updates. Unlike prior soft clipping approaches that use sigmoid-based transformations which can suffer from vanishing gradients and saturation, our method uses a linear interpolation, providing simpler and more robust gradient preservation. Empirically, GR-PSPO outperforms clipping and sigmoid-based alternatives on mathematical reasoning benchmarks when refining models with prior domain knowledge, achieving an accuracy of 79.9% on GSM8K and 59.6% on MATH for Qwen2-Math-1.5B.
Authors: Akhil Premkumar
Abstract: Diffusion models transform noise into data by injecting information that was captured in their neural network during the training phase. In this paper, we ask: \textit{what} is this information? We find that, in pixel-space diffusion models, (1) a large fraction of the total information in the neural network is committed to reconstructing small-scale perceptual details of the image, and (2) the correlations between images and their class labels are informed by the semantic content of the images, and are largely agnostic to the low-level details. We argue that these properties are intrinsically tied to the manifold structure of the data itself. Finally, we show that these facts explain the efficacy of classifier-free guidance: the guidance vector amplifies the mutual information between images and conditioning signals early in the generative process, influencing semantic structure, but tapers out as perceptual details are filled in.
Authors: Jongwook Han, Jongwon Lim, Injin Kong, Yohan Jo
Abstract: Large language models can express values in two main ways: (1) intrinsic expression, reflecting the model's inherent values learned during training, and (2) prompted expression, elicited by explicit prompts. Given their widespread use in value alignment, it is paramount to clearly understand their underlying mechanisms, particularly whether they mostly overlap (as one might expect) or rely on distinct mechanisms, but this remains largely understudied. We analyze this at the mechanistic level using two approaches: (1) value vectors, feature directions representing value mechanisms extracted from the residual stream, and (2) value neurons, MLP neurons that contribute to value vectors. We demonstrate that intrinsic and prompted value mechanisms partly share common components crucial for inducing value expression, generalizing across languages and reconstructing theoretical inter-value correlations in the model's internal representations. Yet, as these mechanisms also possess unique elements that fulfill distinct roles, they lead to different degrees of response diversity (intrinsic > prompted) and value steerability (prompted > intrinsic). In particular, components unique to the intrinsic mechanism promote lexical diversity in responses, whereas those specific to the prompted mechanism strengthen instruction following, taking effect even in distant tasks like jailbreaking.
Authors: Lei Tong, Zhihua Liu, Chaochao Lu, Dino Oglic, Tom Diethe, Philip Teare, Sotirios A. Tsaftaris, Chen Jin
Abstract: We present Causal-Adapter, a modular framework that adapts frozen text-to-image diffusion backbones for counterfactual image generation. Our method enables causal interventions on target attributes, consistently propagating their effects to causal dependents without altering the core identity of the image. In contrast to prior approaches that rely on prompt engineering without explicit causal structure, Causal-Adapter leverages structural causal modeling augmented with two attribute regularization strategies: prompt-aligned injection, which aligns causal attributes with textual embeddings for precise semantic control, and a conditioned token contrastive loss to disentangle attribute factors and reduce spurious correlations. Causal-Adapter achieves state-of-the-art performance on both synthetic and real-world datasets, with up to 91% MAE reduction on Pendulum for accurate attribute control and 87% FID reduction on ADNI for high-fidelity MRI image generation. These results show that our approach enables robust, generalizable counterfactual editing with faithful attribute modification and strong identity preservation.
Authors: Chenhua Shi, Gregor Macdonald, Bhavika Jalli, Wanlu Lei, John Zou, Mridul Jain, Joji Philip
Abstract: The success of large language models (LLMs) depends heavily on large-scale, high-quality instruction-following and reinforcement datasets. However, generating such data through human annotation is prohibitively time-consuming particularly for domain-specific tasks like telecom network troubleshooting, where accurate responses require deep technical expertise and contextual understanding. In this paper, we present a fully automated, retrieval-augmented pipeline for generating synthetic question-answer (QA) pairs grounded in structured domain knowledge. Our multi-stage framework integrates a retriever, base generator, and refinement model to synthesize and enhance QA pairs using documents retrieved from a domain-specific knowledge graph. To ensure data quality, we employ customized RAGAS-based scoring to filter low-quality samples, producing a high-quality dataset suitable for reinforcement fine-tuning (RFT). We demonstrate our approach in a real-world telecom scenario focused on radio access network (RAN) troubleshooting. The resulting pipeline generates complex, context-rich troubleshooting solution plans without human intervention. This work offers a scalable solution for building instruction and reinforcement datasets in specialized domains, significantly reducing dependence on manual labeling while maintaining high technical fidelity.
Authors: Seohyun Lee, Wenzhi Fang, Dong-Jun Han, Seyyedali Hosseinalipour, Christopher G. Brinton
Abstract: In federated learning (FL), local personalization of models has received significant attention, yet personalized fine-tuning of foundation models remains a significant challenge. In particular, there is a lack of understanding in the literature on how to fine-tune and personalize foundation models in settings that are heterogeneous across clients not only in data, but also in tasks and modalities. To address this gap, we propose TAP (Two-Stage Adaptive Personalization), which has two key features: (i) leveraging mismatched model architectures between the clients and server to selectively conduct replacement operations when it benefits a client's local tasks; (ii) engaging in post-FL knowledge distillation for capturing beneficial general knowledge without compromising personalization. In developing TAP, we introduce the first convergence analysis of federated foundation model training at the server under its modality-task pair architecture, and demonstrate that as the number of modality-task pairs increases, its ability to cater to all tasks suffers. Through extensive experiments, we demonstrate the effectiveness of our proposed algorithm across a variety of datasets and tasks in comparison to state-of-the-art federated personalization baselines.
Authors: Houjun Liu, Shikhar Murty, Christopher D. Manning, R\'obert Csord\'as
Abstract: Current approaches for scaling inference-time compute in transformers train them to emit explicit chain-of-thought tokens before producing an answer. While these methods are powerful, they are limited because they cannot be applied during pretraining and rely solely on serially-generated, natural-language verbalization. In this work, we propose Thoughtbubbles, a transformer variant that natively performs parallel adaptive computation in latent space by learning to fork or delete residual streams. Thus, tokens requiring more computation can form a "bubble" of cloned residuals in the middle of the network. Crucially, this behavior is learned during pretraining with only language modeling loss. Using half of the training budget, Thoughtbubbles outperforms the perplexity and zero-shot evals of both standard decoder LMs and those using non-adaptive parallel computation approaches. These results hold across model sizes from 150M to 1.9B. Thoughtbubbles achieves competitive GSM8K results using half of the baseline's token budget. The implicit nature of our method enables models to begin learning adaptive computation at pretraining time, paving the way to unified train-time and test-time scaling behaviors.
Authors: Enxin Song, Wenhao Chai, Shusheng Yang, Ethan Armand, Xiaojun Shan, Haiyang Xu, Jianwen Xie, Zhuowen Tu
Abstract: Video understanding in multimodal language models remains limited by context length: models often miss key transition frames and struggle to maintain coherence across long time scales. To address this, we adapt Native Sparse Attention (NSA) to video-language models. Our method, VideoNSA, adapts Qwen2.5-VL through end-to-end training on a 216K video instruction dataset. We employ a hardware-aware hybrid approach to attention, preserving dense attention for text, while employing NSA for video. Compared to token-compression and training-free sparse baselines, VideoNSA achieves improved performance on long-video understanding, temporal reasoning, and spatial benchmarks. Further ablation analysis reveals four key findings: (1) reliable scaling to 128K tokens; (2) an optimal global-local attention allocation at a fixed budget; (3) task-dependent branch usage patterns; and (4) the learnable combined sparse attention help induce dynamic attention sinks. Project Page: https://enxinsong.com/VideoNSA-web/, Code: https://github.com/Espere-1119-Song/VideoNSA
URLs: https://enxinsong.com/VideoNSA-web/,, https://github.com/Espere-1119-Song/VideoNSA
Authors: Xianglong Yan, Chengzhu Bao, Zhiteng Li, Tianao Zhang, Kaicheng Yang, Haotong Qin, Ruobing Xie, Xingwu Sun, Yulun Zhang
Abstract: Large Language Models (LLMs) have shown impressive capabilities across diverse tasks, but their large memory and compute demands hinder deployment. Ternarization has gained attention as a promising compression technique, delivering substantial size reduction and high computational efficiency. However, its potential in the post-training quantization (PTQ) setting remains underexplored, due to the challenge of training-free parameter optimization and the quantization difficulty posed by outliers and dispersed weights. To address these issues, we propose PT$^2$-LLM, a post-training ternarization framework tailored for LLMs. At its core is an Asymmetric Ternary Quantizer equipped with a two-stage refinement pipeline: (1) Iterative Ternary Fitting (ITF), which alternates between optimal ternary grid construction and flexible rounding to minimize quantization error, and (2) Activation-aware Grid Alignment (AGA), which further refines the ternary grid to better match full-precision outputs. In addition, we propose a plug-and-play Structural Similarity-based Reordering (SSR) strategy that leverages inter-column structural similarity to ease quantization and mitigate outlier effects, further enhancing overall performance. Extensive experiments demonstrate that PT$^2$-LLM delivers competitive performance against state-of-the-art (SOTA) 2-bit PTQ methods with lower memory cost, while also accelerating both prefill and decoding to achieve end-to-end speedup. The code and models will be available at https://github.com/XIANGLONGYAN/PT2-LLM.
Authors: Junde Xu, Yapin Shi, Lijun Lang, Taoyong Cui, Zhiming Zhang, Guangyong Chen, Jiezhong Qiu, Pheng-Ann Heng
Abstract: Multimodal protein language models deliver strong performance on mutation-effect prediction, but training such models from scratch demands substantial computational resources. In this paper, we propose a fine-tuning framework called InstructPLM-mu and try to answer a question: \textit{Can multimodal fine-tuning of a pretrained, sequence-only protein language model match the performance of models trained end-to-end? } Surprisingly, our experiments show that fine-tuning ESM2 with structural inputs can reach performance comparable to ESM3. To understand how this is achieved, we systematically compare three different feature-fusion designs and fine-tuning recipes. Our results reveal that both the fusion method and the tuning strategy strongly affect final accuracy, indicating that the fine-tuning process is not trivial. We hope this work offers practical guidance for injecting structure into pretrained protein language models and motivates further research on better fusion mechanisms and fine-tuning protocols.
Authors: Qizheng Zhang, Changran Hu, Shubhangi Upasani, Boyuan Ma, Fenglu Hong, Vamsidhar Kamanuru, Jay Rainton, Chen Wu, Mengmeng Ji, Hanchen Li, Urmish Thakker, James Zou, Kunle Olukotun
Abstract: Large language model (LLM) applications such as agents and domain-specific reasoning increasingly rely on context adaptation -- modifying inputs with instructions, strategies, or evidence, rather than weight updates. Prior approaches improve usability but often suffer from brevity bias, which drops domain insights for concise summaries, and from context collapse, where iterative rewriting erodes details over time. Building on the adaptive memory introduced by Dynamic Cheatsheet, we introduce ACE (Agentic Context Engineering), a framework that treats contexts as evolving playbooks that accumulate, refine, and organize strategies through a modular process of generation, reflection, and curation. ACE prevents collapse with structured, incremental updates that preserve detailed knowledge and scale with long-context models. Across agent and domain-specific benchmarks, ACE optimizes contexts both offline (e.g., system prompts) and online (e.g., agent memory), consistently outperforming strong baselines: +10.6% on agents and +8.6% on finance, while significantly reducing adaptation latency and rollout cost. Notably, ACE could adapt effectively without labeled supervision and instead by leveraging natural execution feedback. On the AppWorld leaderboard, ACE matches the top-ranked production-level agent on the overall average and surpasses it on the harder test-challenge split, despite using a smaller open-source model. These results show that comprehensive, evolving contexts enable scalable, efficient, and self-improving LLM systems with low overhead.
Authors: P\'al Zs\'amboki, Benjamin Levi, David Ansel Josef Smith, Mitansh Kagalwala, Arlington Kell, Samuel Liechty, Cong Wang
Abstract: Length Generalization is the essential capacity of autonomous agents to perform tasks in longer contexts than those encountered during training. To systematically study this feat, we test how well models can approximate the next token distributions in algorithmic tasks. This is to take into account the realistic possibility of multiple next tokens being legal. We present a prototypical benchmark for this line of study: in the Set Complement Task, the model needs to output a uniform distribution over tokens not in the input. We prove a theorem that states simple transformers can length generalize on this task, however, with performance degradation due to attention dispersion. A mechanistic reading of how dispersion takes effect lets us discover a remedy: Post-Norm can Resharpen Attention. We present experimental evidence to support this idea. We also show that Exponential Moving Averages can help the issue of noisy gradients that arises when many next tokens are legal. We validate the general applicability of our proposed methods on a suite of formal language experiments. Our source code will be available upon publication.
Authors: Tilman Hinnerichs, Reuben Gardos Reid, Jaap de Jong, Bart Swinkels, Pamela Wochner, Nicolae Filat, Tudor Magurescu, Issa Hanou, Sebastijan Dumancic
Abstract: Program synthesis -- the automatic generation of code given a specification -- is one of the most fundamental tasks in artificial intelligence (AI) and the dream of many programmers. Numerous synthesizers have been developed for program synthesis, offering different approaches to the exponentially growing program space. Although such state-of-the-art tools exist, reusing and adapting them remains tedious and time-consuming. We propose Herb.jl, a unifying program synthesis library written in Julia, to address these issues. Since current methods share similar building blocks, we aim to break down the underlying algorithms into extendable, reusable subcomponents. To demonstrate the benefits of using Herb.jl, we show how to implement a simple problem and grammar, and how to solve it with just a few lines of code.
Authors: Zongyi Lyu, Songqiang Chen, Zhenlan Ji, Liwen Wang, Shuai Wang, Daoyuan Wu, Wenxuan Wang, Shing-Chi Cheung
Abstract: Multi-agent systems (MASs) have emerged as a promising paradigm for automated code generation, demonstrating impressive performance on established benchmarks. Despite their prosperous development, the fundamental mechanisms underlying their robustness remain poorly understood, raising critical concerns for real-world deployment. This paper conducts a systematic empirical study to uncover the internal robustness flaws of MASs using a mutation-based methodology. By designing a testing pipeline incorporating semantic-preserving mutation operators and a novel fitness function, we assess mainstream MASs across multiple datasets and LLMs. Our findings reveal substantial robustness flaws: semantically equivalent inputs cause drastic performance drops, with MASs failing to solve 7.9\%--83.3\% of problems they initially resolved successfully. Through comprehensive failure analysis, we discover a fundamental cause underlying these robustness issues: the \textit{planner-coder gap}, which accounts for 75.3\% of failures. This gap arises from information loss in the multi-stage transformation process where planning agents decompose requirements into underspecified plans, and coding agents subsequently misinterpret intricate logic during code generation. Based on this formulated information transformation process, we propose a \textit{repairing method} that mitigates information loss through multi-prompt generation and introduces a monitor agent to bridge the planner-coder gap. Evaluation shows that our repairing method effectively enhances the robustness of MASs by solving 40.0\%--88.9\% of identified failures. Our work uncovers critical robustness flaws in MASs and provides effective mitigation strategies, contributing essential insights for developing more reliable MASs for code generation.
Authors: Nicolas Menet, Aleksandar Terzi\'c, Michael Hersche, Andreas Krause, Abbas Rahimi
Abstract: Bayesian optimization in large unstructured discrete spaces is often hindered by the computational cost of maximizing acquisition functions due to the absence of gradients. We propose a scalable alternative based on Thompson sampling that eliminates the need for acquisition function maximization by directly parameterizing the probability that a candidate yields the maximum reward. Our approach, Thompson Sampling via Fine-Tuning (ToSFiT) leverages the prior knowledge embedded in prompt-conditioned large language models, and incrementally adapts them toward the posterior. Theoretically, we derive a novel regret bound for a variational formulation of Thompson Sampling that matches the strong guarantees of its standard counterpart. Our analysis reveals the critical role of careful adaptation to the posterior probability of maximality -- a principle that underpins our ToSFiT algorithm. Empirically, we validate our method on three diverse tasks: FAQ response refinement, thermally stable protein search, and quantum circuit design. Within a collection of methods covering Bayesian optimization, reinforcement learning, and evolutionary search, ToSFiT exhibits both state-of-the-art sample efficiency and computational efficiency.
Authors: David Atkinson
Abstract: Open-source status should not shield generative artificial intelligence systems from ethical or legal accountability. Through a rigorous analysis of regulatory, legal, and policy frameworks, this Article contends that open-source GenAI must be held to the same standards as proprietary systems. While recognizing the value of openness for scientific advancement, I propose a narrowly tailored safe harbor for bona fide, non-commercial research, conditioned on strict compliance with defined criteria. This Article critically examines and refutes the core claims of open-source exceptionalism--namely, that open-source GenAI disrupts entrenched oligopolies, democratizes access, and uniquely drives innovation. The evidence shows that open-source GenAI can facilitate unlawful conduct, exacerbate environmental harms, and reinforce existing power structures. Rhetoric around "democratization" and "innovation" often serves as an unsubstantiated basis for regulatory exemptions not afforded to proprietary systems. This Article ultimately advocates for a framework that promotes responsible AI development, balancing openness with robust legal and ethical safeguards and a clear-eyed assessment of societal impacts.
Authors: David Atkinson
Abstract: GenAI companies are strip-mining the web. Their scraping bots harvest content at an unprecedented scale, circumventing technical barriers to fuel billion-dollar models while creators receive nothing. Courts have enabled this exploitation by misunderstanding what property rights protect online. The prevailing view treats websites as mere repositories of intellectual property and dismisses trespass claims absent server damage. That framework grants AI companies presumptive access while ignoring the economic devastation they inflict. But the content is severable from the website itself. This paper reframes the debate: websites are personal property as integrated digital assets subject to the same exclusionary rights as physical chattels. When scrapers bypass access controls and divert traffic that sustains a website's value, they commit actionable trespass. The law need not create new protections; it need only apply existing property principles to digital space. Courts and litigants have struggled to police unwanted, large-scale scraping because copyright preemption often narrows available claims, leaving copyright and its fair use defense as the primary battleground. Trespass to chattels offers a superior path, grounded in the fundamental right to exclude unwanted intrusions. Reviving this tort would protect not only content creators but also the digital ecosystem. Such protection would discourage exploitative scraping, preserve incentives for content creation, help protect privacy and personal data, and safeguard autonomy and expression. Reaffirming website owners' right to exclude is essential to maintaining a fair and sustainable online environment.
Authors: Peihong Zhang, Yuxuan Liu, Rui Sang, Zhixin Li, Yiqiang Cai, Yizhou Tan, Shengchen Li
Abstract: Acoustic scene classification (ASC) suffers from device-induced domain shift, especially when labels are limited. Prior work focuses on curriculum-based training schedules that structure data presentation by ordering or reweighting training examples from easy-to-hard to facilitate learning; however, existing curricula are static, fixing the ordering or the weights before training and ignoring that example difficulty and marginal utility evolve with the learned representation. To overcome this limitation, we propose the Dynamic Dual-Signal Curriculum (DDSC), a training schedule that adapts the curriculum online by combining two signals computed each epoch: a domain-invariance signal and a learning-progress signal. A time-varying scheduler fuses these signals into per-example weights that prioritize domain-invariant examples in early epochs and progressively emphasize device-specific cases. DDSC is lightweight, architecture-agnostic, and introduces no additional inference overhead. Under the official DCASE 2024 Task~1 protocol, DDSC consistently improves cross-device performance across diverse ASC baselines and label budgets, with the largest gains on unseen-device splits.
Authors: Peihong Zhang, Zhixin Li, Yuxuan Liu, Rui Sang, Yiqiang Cai, Yizhou Tan, Shengchen Li
Abstract: Deep learning approaches for heart-sound (PCG) segmentation built on time-frequency features can be accurate but often rely on large expert-labeled datasets, limiting robustness and deployment. We present TopSeg, a topological representation-centric framework that encodes PCG dynamics with multi-scale topological features and decodes them using a lightweight temporal convolutional network (TCN) with an order- and duration-constrained inference step. To evaluate data efficiency and generalization, we train exclusively on PhysioNet 2016 dataset with subject-level subsampling and perform external validation on CirCor dataset. Under matched-capacity decoders, the topological features consistently outperform spectrogram and envelope inputs, with the largest margins at low data budgets; as a full system, TopSeg surpasses representative end-to-end baselines trained on their native inputs under the same budgets while remaining competitive at full data. Ablations at 10% training confirm that all scales contribute and that combining H_0 and H_1 yields more reliable S1/S2 localization and boundary stability. These results indicate that topology-aware representations provide a strong inductive bias for data-efficient, cross-dataset PCG segmentation, supporting practical use when labeled data are limited.
Authors: Precious Eze (College of Engineering,Computing, Florida International University, Miami, USA), Stephanie Lunn (College of Engineering,Computing, Florida International University, Miami, USA), Bruk Berhane (College of Engineering,Computing, Florida International University, Miami, USA)
Abstract: Employers increasingly expect graduates to utilize large language models (LLMs) in the workplace, yet the competencies needed for computing roles across Africa remain unclear given varying national contexts. This study examined how six LLMs, namely ChatGPT 4, DeepSeek, Gemini, Claude 3.5, Llama 3, and Mistral AI, describe entry-level computing career expectations across ten African countries. Using the Computing Curricula 2020 framework and drawing on Digital Colonialism Theory and Ubuntu Philosophy, content analysis of 60 LLM responses to standardized prompts reveals consistent coverage of technical competencies such as cloud computing and programming, but notable differences in non-technical competencies, particularly ethics and responsible AI use. Models vary considerably in recognizing country-specific factors, including local technology ecosystems, language requirements, and national policies averaging only 35.4% contextual awareness overall. Open-source models demonstrated stronger contextual awareness and better balance between technical and professional skills, with Llama (4.47/5) and DeepSeek (4.25/5) outperforming proprietary alternatives ChatGPT-4 (3.90/5) and Claude (3.46/5). However, Mistral's poor contextual performance (0.00/4) despite being open-source indicates that development philosophy alone does not guarantee contextual responsiveness. This first comprehensive comparison of LLM career guidance for African computing students uncovers entrenched infrastructure assumptions and Western-centric biases that create gaps between technical recommendations and local realities. The findings challenge assumptions about AI tool quality in resource-constrained settings and underscore the need for decolonial approaches to AI in education, emphasizing contextual relevance and hybrid human-AI guidance models.
Authors: Afrozah Nadeem, Mark Dras, Usman Naseem
Abstract: Large language models often display undesirable behaviors embedded in their internal representations, undermining fairness, inconsistency drift, amplification of harmful content, and the propagation of unwanted patterns during extended dialogue and conversations. Although training-time or data-centric methods attempt to reduce these effects, they are computationally expensive, irreversible once deployed, and slow to adapt to new conversational contexts. Pruning-based methods provide a flexible and transparent way to reduce bias by adjusting the neurons responsible for certain behaviors. However, most existing approaches are static; once a neuron is removed, the model loses the ability to adapt when the conversation or context changes. To address this, we propose a dynamic, reversible, pruning-based framework that detects context-aware neuron activations and applies adaptive masking to modulate their influence during generation. Our inference-time solution provides fine-grained, memory-aware mitigation with knowledge-preserved, more coherent behavior across multilingual single- and multi-turn dialogues, enabling dynamic fairness control in real-world conversational AI.
Authors: Guilin Zhang, Wulan Guo, Ziqi Tan, Srinivas Vippagunta, Suchitra Raman, Shreeshankar Chatterjee, Ju Lin, Shang Liu, Mary Schladenhauffen, Jeffrey Luo, Hailong Jiang
Abstract: Industrial and government organizations increasingly depend on data-driven analytics for workforce, finance, and regulated decision processes, where timeliness, cost efficiency, and compliance are critical. Distributed frameworks such as Spark and Flink remain effective for massive-scale batch or streaming analytics but introduce coordination complexity and auditing overheads that misalign with moderate-scale, latency-sensitive inference. Meanwhile, cloud providers now offer serverless GPUs, and models such as TabNet enable interpretable tabular ML, motivating new deployment blueprints for regulated environments. In this paper, we present a production-oriented Big Data as a Service (BDaaS) blueprint that integrates a single-node serverless GPU runtime with TabNet. The design leverages GPU acceleration for throughput, serverless elasticity for cost reduction, and feature-mask interpretability for IL4/FIPS compliance. We conduct benchmarks on the HR, Adult, and BLS datasets, comparing our approach against Spark and CPU baselines. Our results show that GPU pipelines achieve up to 4.5x higher throughput, 98x lower latency, and 90% lower cost per 1K inferences compared to Spark baselines, while compliance mechanisms add only ~5.7 ms latency with p99 < 22 ms. Interpretability remains stable under peak load, ensuring reliable auditability. Taken together, these findings provide a compliance-aware benchmark, a reproducible Helm-packaged blueprint, and a decision framework that demonstrate the practicality of secure, interpretable, and cost-efficient serverless GPU analytics for regulated enterprise and government settings.
Authors: Xuhui Zhou, Valerie Chen, Zora Zhiruo Wang, Graham Neubig, Maarten Sap, Xingyao Wang
Abstract: Recent advances in coding agents have made them capable of planning, editing, running, and testing complex code bases. Despite their growing ability in coding tasks, these systems still struggle to infer and track user intent, especially when instructions are underspecified or context-dependent. To bridge this gap, we introduce ToM-SWE, a dual-agent architecture that pairs a primary software-engineering (SWE) agent with a lightweight theory-of-mind (ToM) partner agent dedicated to modeling the user's mental state. The ToM agent infers user goals, constraints, and preferences from instructions and interaction history, maintains a \textbf{persistent memory} of the user, and provides user-related suggestions to the SWE agent. In two software engineering benchmarks (ambiguous SWE-bench and stateful SWE-bench), ToM-SWE improves task success rates and user satisfaction. Notably, on the stateful SWE benchmark, a newly introduced evaluation that provides agents with a user simulator along with previous interaction histories, ToM-SWE achieves a substantially higher task success rate of 59.7\% compared to 18.1\% for OpenHands, a state-of-the-art SWE agent. Furthermore, in a three-week study with professional developers using ToM-SWE in their daily work, participants found it useful 86\% of the time, underscoring the value of stateful user modeling for practical coding agents.
Authors: Benjamin Reichman, Adar Avsian, Larry Heck
Abstract: This work investigates how large language models (LLMs) internally represent emotion by analyzing the geometry of their hidden-state space. The paper identifies a low-dimensional emotional manifold and shows that emotional representations are directionally encoded, distributed across layers, and aligned with interpretable dimensions. These structures are stable across depth and generalize to eight real-world emotion datasets spanning five languages. Cross-domain alignment yields low error and strong linear probe performance, indicating a universal emotional subspace. Within this space, internal emotion perception can be steered while preserving semantics using a learned intervention module, with especially strong control for basic emotions across languages. These findings reveal a consistent and manipulable affective geometry in LLMs and offer insight into how they internalize and process emotion.
Authors: Anchit Jain, Stephen Bates
Abstract: Decomposing prediction uncertainty into aleatoric (irreducible) and epistemic (reducible) components is critical for the reliable deployment of machine learning systems. While the mutual information between the response variable and model parameters is a principled measure for epistemic uncertainty, it requires access to the parameter posterior, which is computationally challenging to approximate. Consequently, practitioners often rely on probabilistic predictions from deep ensembles to quantify uncertainty, which have demonstrated strong empirical performance. However, a theoretical understanding of their success from a frequentist perspective remains limited. We address this gap by first considering a bootstrap-based estimator for epistemic uncertainty, which we prove is asymptotically correct. Next, we connect deep ensembles to the bootstrap estimator by decomposing it into data variability and training stochasticity; specifically, we show that deep ensembles capture the training stochasticity component. Through empirical studies, we show that this stochasticity component constitutes the majority of epistemic uncertainty, thereby explaining the effectiveness of deep ensembles.
Authors: Ranran Haoran Zhang, Soumik Dey, Ashirbad Mishra, Hansi Wu, Binbin Li, Rui Zhang
Abstract: Speculative decoding must produce outputs distribution identical to standard autoregressive generation-this output equivalence is not an optimization target but the defining criterion of valid speculative decoding. We demonstrate that all existing batch speculative decoding implementations violate this fundamental requirement, producing corrupted outputs ranging from repetitive tokens to gibberish. These failures stem from the ragged tensor problem: sequences in the same batch accept different numbers of draft tokens, desynchronizing position IDs, attention masks, and KV-cache state. We present the first authentic batch speculative decoding framework. We (1) formalize the synchronization invariants that valid batch speculative decoding must satisfy, (2) present EQSPEC, the first algorithm that guarantees output equivalence, and analyze its cost structure to show that alignment overhead grows superlinearly and consumes up to 40\% of computation, and (3) introduce EXSPEC, which reduces this overhead through cross-batch scheduling that dynamically groups same-length sequences. On SpecBench across Vicuna-7B/68M, Qwen3-8B/0.6B, and GLM-4-9B/0.6B pairs, our methods achieve up to 3x throughput improvement at batch size 8 while maintaining algorithmic correctness. Our methods achieve 95\% decoding-equivalence, with residual divergence attributable to floating-point non-determinism in GPU inference, not the synchronization failures that cause near-zero equivalence of prior methods. Our code is available at https://github.com/eBay/spec_dec.
Authors: Marko Karbevski, Antonij Mijoski
Abstract: We theoretically investigate whether the Query, Key, Value weight triplet can be reduced in encoder-only and decoder-only transformers. Under mild assumptions, we prove that Query weights are redundant and can be replaced with the identity matrix, reducing attention parameters by $25\%$. This also simplifies optimization: attention logits become linear rather than quadratic in learned weights. Validating on decoder-only GPT-style small models trained from scratch, we find that with adjusted attention scaling and weight decay, reduced models match baseline performance despite fewer parameters. Training remains stable at over $3\times$ lower weight decay, suggesting Query weight elimination provides implicit regularization. Our analysis has also led us to a structural expressivity boundary: in the mathematically tractable ReLU setting, skip connections push MLPs into a generically disjoint function class at fixed width. These findings motivate investigation across modalities and at scale, where the observed stability and efficiency gains may prove most consequential.
Authors: Kun Chen, Peng Shi, Haibo Qiu, Zhixiong Zeng, Siqi Yang, Wenji Mao, Lin Ma
Abstract: Reinforcement learning (RL) with verifiable rewards has recently catalyzed a wave of "MLLM-r1" approaches that bring RL to vision language models. Most representative paradigms begin with a cold start, typically employing supervised fine-tuning (SFT), to initialize the policy before RL. However, SFT-based cold start adopts the reasoning paradigm intertwined with task solution and output format, which may induce instruction-style overfitting, weakens out-of-distribution generalization, and ultimately affects downstream RL. We revisit the cold start along two views, its training method and data construction, and introduce the Generalization Factor (GF) coefficient to quantify the generalization capability under different methods. Our empirical study finds that preference-based training methods (e.g. DPO) generalizes better than SFT-based methods in cold start. Motivated by this, we propose SPECS-a Self-distilled, Preference-based Cold Start framework that decouples multimodal learning: (1) generates introspective preference data pairs via self-distillation, avoiding reliance on larger teachers or manual annotation; (2) performs preference-based training to learn, focusing on shallow, transferable surface-form criteria (format, structure, style) rather than memorizing content; and (3) hands off to RL with verifiable rewards for deep reasoning results. Experimental results across multiple multimodal benchmarks show that our decoupling learning framework yields consistent performance gains over strong baselines, improving MEGA-Bench by 4.1% and MathVista by 12.2%. Additional experiments indicate that SPECS contributes to reducing in-distribution "stuckness," improving exploration, stabilizing training, and raising the performance ceiling. Project Page: https://kwen-chen.github.io/SPECS-VL/
Authors: Xiangqing Zheng, Chengyue Wu, Kehai Chen, Min Zhang
Abstract: Recent advances in text-to-video generation have achieved impressive performance on short clips, yet evaluating long-form generation under complex textual inputs remains a significant challenge. In response to this challenge, we present LoCoT2V-Bench, a benchmark for long video generation (LVG) featuring multi-scene prompts with hierarchical metadata (e.g., character settings and camera behaviors), constructed from collected real-world videos. We further propose LoCoT2V-Eval, a multi-dimensional framework covering perceptual quality, text-video alignment, temporal quality, dynamic quality, and Human Expectation Realization Degree (HERD), with an emphasis on aspects such as fine-grained text-video alignment and temporal character consistency. Experiments on 13 representative LVG models reveal pronounced capability disparities across evaluation dimensions, with strong perceptual quality and background consistency but markedly weaker fine-grained text-video alignment and character consistency. These findings suggest that improving prompt faithfulness and identity preservation remains a key challenge for long-form video generation.
Authors: Zhixing Li, Arsham Gholamzadeh Khoee, Yinan Yu
Abstract: The objective of domain generalization (DG) is to enable models to be robust against domain shift. DG is crucial for deploying vision-language models (VLMs) in real-world applications, yet most existing methods rely on domain labels that may not be available and often ambiguous. We instead study the DG setting where models must generalize well without access to explicit domain labels. Our key idea is to represent an unseen target domain as a combination of latent domains automatically discovered from training data, enabling the model to adaptively transfer knowledge across domains. To realize this, we perform latent domain clustering on image features and fuse domain-specific text features based on the similarity between the input image and each latent domain. Experiments on four benchmarks show that this strategy yields consistent gains over VLM-based baselines and provides new insights into improving robustness under domain shift.
Authors: Sekh Mainul Islam, Pepa Atanasova, Isabelle Augenstein
Abstract: Natural Language Explanations (NLEs) describe how Large Language Models (LLMs) make decisions by drawing on external Context Knowledge (CK) and Parametric Knowledge (PK). Understanding the interaction between these sources is key to assessing NLE grounding, yet these dynamics remain underexplored. Prior work has largely focused on (1) single-step generation and (2) modelled PK-CK interaction as a binary choice within a rank-1 subspace. This approach overlooks richer interactions and how they unfold over longer generations, such as complementary or supportive knowledge. We propose a novel rank-2 projection subspace that disentangles PK and CK contributions more accurately and use it for the first multi-step analysis of knowledge interactions across longer NLE sequences. Experiments across four QA datasets and three open-weight LLMs demonstrate that while rank-1 subspaces struggle to represent diverse interactions, our rank-2 formulation captures them effectively, highlighting PK alignment for supportive interactions and CK alignment for conflicting ones. Our multi-step analysis reveals, among others, that hallucinated generations exhibit strong alignment with the PK direction, whereas context-faithful generations maintain a more balanced alignment between PK and CK.
Authors: Hanchen Li, Qiuyang Mang, Runyuan He, Qizheng Zhang, Huanzhi Mao, Xiaokun Chen, Hangrui Zhou, Alvin Cheung, Joseph Gonzalez, Ion Stoica
Abstract: KV cache management is essential for efficient LLM inference. To maximize utilization, existing inference engines evict finished requests' KV cache if new requests are waiting. This policy breaks for agentic workloads, which interleave LLM calls with tools, introducing pauses that prevent effective KV reuse across turns. Since some tool calls have much shorter durations than human response multi-turn chatbot, it would be promising to retain the KV cache in during these tools. However, there are many challenges. First, we need to consider both the potential cost of recomputation or reloading (if CPU offloading enabled) and the increasing queueing delays after eviction from GPU. Second, due to the internal variance of tool call durations, we need the method to remain robust under limited predictability of tool call durations. We present Continuum, a serving system to optimize job completion time for multi-turn agent workloads by introducing time-to-live mechanism for KV cache retaining. For LLM request that generates a tool call, Continuum selectively pins the KV cache in GPU memory with a time-to-live value determined by considering both the reload cost and ordering preserve benefit of retaining KV cache. Moreover, when the TTL expires, the KV cache can be automatically evicted to free up GPU memory, providing robust performance under edge cases. When combined with program-level first-come-first-serve, Continuum preserves multi-turn continuity, and reduces delay for complex agentic workflows. Our evaluation on real-world agentic workloads (SWE-Bench and BFCL) with Llama-3.1 8B/70B shows that Continuum significantly improves the average job completion times and its improvement scales with turn number increase. We release a preview version at: https://github.com/Hanchenli/vllm-continuum
Authors: Dongsu Lee, Daehee Lee, Amy Zhang
Abstract: This work presents MAC-Flow, a simple yet expressive framework for multi-agent coordination. We argue that requirements of effective coordination are twofold: (i) a rich representation of the diverse joint behaviors present in offline data and (ii) the ability to act efficiently in real time. However, prior approaches often sacrifice one for the other, i.e., denoising diffusion-based solutions capture complex coordination but are computationally slow, while Gaussian policy-based solutions are fast but brittle in handling multi-agent interaction. MAC-Flow addresses this trade-off by first learning a flow-based representation of joint behaviors, and then distilling it into decentralized one-step policies that preserve coordination while enabling fast execution. Across four different benchmarks, including $12$ environments and $34$ datasets, MAC-Flow alleviates the trade-off between performance and computational cost, specifically achieving about $\boldsymbol{\times14.5}$ faster inference compared to diffusion-based MARL methods, while maintaining good performance. At the same time, its inference speed is similar to that of prior Gaussian policy-based offline multi-agent reinforcement learning (MARL) methods.
Authors: Zijian Chen, Wenjun Zhang, Guangtao Zhai
Abstract: The potential data contamination issue in contemporary large language models (LLMs) benchmarks presents a fundamental challenge to establishing trustworthy evaluation frameworks. Meanwhile, they predominantly assume benign, resource-rich settings, leaving the behavior of LLMs under pressure unexplored. In this paper, we introduce \textsc{Squid Game}, a dynamic and adversarial evaluation environment with resource-constrained and asymmetric information settings elaborated to evaluate LLMs through interactive gameplay against other LLM opponents. Squid Game consists of six elimination-style levels, focusing on multi-faceted abilities, including instruction-following, code, reasoning, planning, and safety alignment. We evaluate over 50 LLMs on Squid Game, presenting the largest behavioral evaluation study of general LLMs on dynamic adversarial scenarios. We observe a clear generational phase transition in performance in the same model lineage and find evidence that some models resort to speculative shortcuts to win the game, indicating the possibility of higher-level evaluation paradigm contamination in static benchmarks. We also compare prominent LLM benchmarks and \textsc{Squid Game}, highlighting that dynamic evaluation can serve as a complementary part for static evaluations. Project page: https://github.com/zijianchen98/LLM_Squid_Game.
Authors: Sechan Lee, Sangdon Park
Abstract: Recently, web-based Large Language Model (LLM) agents autonomously perform increasingly complex tasks, thereby bringing significant convenience. However, they also amplify the risks of malicious misuse cases such as unauthorized collection of personally identifiable information (PII), generation of socially divisive content, and even automated web hacking. To address these threats, we propose an AI Kill Switch technique that can immediately halt the operation of malicious web-based LLM agents. To achieve this, we introduce AutoGuard - the key idea is generating defensive prompts that trigger the safety mechanisms of malicious LLM agents. In particular, generated defense prompts are transparently embedded into the website's DOM so that they remain invisible to human users but can be detected by the crawling process of malicious agents, triggering its internal safety mechanisms to abort malicious actions once read. To evaluate our approach, we constructed a dedicated benchmark consisting of three representative malicious scenarios. Experimental results show that AutoGuard achieves over 80% Defense Success Rate (DSR) across diverse malicious agents, including GPT-4o, Claude-4.5-Sonnet and generalizes well to advanced models like GPT-5.1, Gemini-2.5-flash, and Gemini-3-pro. Also, our approach demonstrates robust defense performance in real-world website environments without significant performance degradation for benign agents. Through this research, we demonstrate the controllability of web-based LLM agents, thereby contributing to the broader effort of AI control and safety.
Authors: Duo Zhou, Yuji Zhang, Tianxin Wei, Ruizhong Qiu, Ke Yang, Xiao Lin, Cheng Qian, Jingrui He, Hanghang Tong, Heng Ji, Huan Zhang
Abstract: Large language models (LLMs) can internalize private or harmful content, motivating unlearning that removes a forget set while preserving retaining knowledge. However, forgetting updates often cause collateral degradation on retaining knowledge, creating a persistent trade-off. Existing LLM unlearning methods are often heuristic, and other theoretical approaches rely on offline feature constructions that do not capture update-time forget-retain interaction in LLMs. To address this limitation, we aim to develop an LLM unlearning method that reduces the forget-retain trade-off with theoretical guarantees. We take a first-principles view by formalizing "no side effects" as local retain invariance under small parameter updates, and prove an equivalence under optimizer-induced geometry: the retain loss is locally invariant if and only if the update direction is orthogonal to the subspace spanned by retain gradients. Based on the insight, we propose Geometric-disentanglement Unlearning (GU), a lightweight and theoretically grounded projection that can be plug-and-play to existing gradient-based unlearning methods to mitigate forget-retain side effects. Experiments on TOFU, MUSE, and WMDP-cyber show that GU strengthens forgetting while reducing retain drift. When added to SimNPO, it achieves up to 62\% improved forgetting Extraction Strength (ES) and 31\% higher retain ES. We open-sourced our code in https://github.com/Lemutisme/Geometric-Unlearning.
Authors: Kirti Bhagat, Shaily Bhatt, Athul Velagapudi, Aditya Vashistha, Shachi Dave, Danish Pruthi
Abstract: Millions of users across the globe turn to AI chatbots for their creative needs, inviting widespread interest in understanding how they represent diverse cultures. However, evaluating cultural representations in open-ended tasks remains challenging and underexplored. In this work, we present TALES, an evaluation of cultural misrepresentations in LLM-generated stories for diverse Indian cultural identities. First, we develop TALES-Tax, a taxonomy of cultural misrepresentations by collating insights from participants with lived experiences in India through focus groups (N=9) and individual surveys (N=15). Using TALES-Tax, we evaluate 6 models through a large-scale annotation study spanning 2925 annotations from 108 annotators with lived experience and native language proficiency from across 71 regions in India and 14 languages. Concerningly, we find that 88% of the generated stories contain misrepresentations, and such errors are more prevalent in mid- and low-resourced languages and stories based in peri-urban regions in India. We also transform the annotations into TALES-QA, a standalone question bank to evaluate the cultural knowledge of models.
Authors: Wei Yang, Rui Zhong, Yiqun Chen, Chi Lu, Peng Jiang
Abstract: Multimodal recommendation aims to integrate collaborative signals with heterogeneous content such as visual and textual information, but remains challenged by modality-specific noise, semantic inconsistency, and unstable propagation over user-item graphs. These issues are often exacerbated by naive fusion or shallow modeling strategies, leading to degraded generalization and poor robustness. While recent work has explored the frequency domain as a lens to separate stable from noisy signals, most methods rely on static filtering or reweighting, lacking the ability to reason over spectral structure or adapt to modality-specific reliability. To address these challenges, we propose a Structured Spectral Reasoning (SSR) framework for frequency-aware multimodal recommendation. Our method follows a four-stage pipeline: (i) Decompose graph-based multimodal signals into spectral bands via graph-guided transformations to isolate semantic granularity; (ii) Modulate band-level reliability with spectral band masking, a training-time masking with a prediction-consistency objective that suppresses brittle frequency components; (iii) Fuse complementary frequency cues using hyperspectral reasoning with low-rank cross-band interaction; and (iv) Align modality-specific spectral features via contrastive regularization to promote semantic and structural consistency. Experiments on three real-world benchmarks show consistent gains over strong baselines, particularly under sparse and cold-start settings. Additional analyses indicate that structured spectral modeling improves robustness and provides clearer diagnostics of how different bands contribute to performance.
Authors: Ruoxin Chen, Jiahui Gao, Kaiqing Lin, Keyue Zhang, Yandan Zhao, Isabel Guan, Taiping Yao, Shouhong Ding
Abstract: Vision Language Models (VLMs) are increasingly used for detecting AI-generated images (AIGI). However, converting VLMs into reliable detectors is resource-intensive, and the resulting models often suffer from hallucination and poor generalization. To investigate the root cause, we conduct an empirical analysis and identify two consistent behaviors. First, fine-tuning VLMs with semantic supervision improves semantic discrimination and generalizes well to unseen data. Second, fine-tuning VLMs with pixel-artifact supervision leads to weak generalization. These findings reveal a fundamental task-model misalignment. VLMs are optimized for high-level semantic reasoning and lack inductive bias toward low-level pixel artifacts. In contrast, conventional vision models effectively capture pixel-level artifacts but are less sensitive to semantic inconsistencies. This indicates that different models are naturally suited to different subtasks. Based on this insight, we formulate AIGI detection as two orthogonal subtasks: semantic consistency checking and pixel-artifact detection. Neglecting either subtask leads to systematic detection failures. We further propose the Task-Model Alignment principle and instantiate it in a two-branch detector, AlignGemini. The detector combines a VLM trained with pure semantic supervision and a vision model trained with pure pixel-artifact supervision. By enforcing clear specialization, each branch captures complementary cues. Experiments on in-the-wild benchmarks show that AlignGemini improves average accuracy by 9.5 percent using simplified training data. These results demonstrate that task-model alignment is an effective principle for generalizable AIGI detection.
Authors: Yuchuan Tian, Yuchen Liang, Shuo Zhang, Yingte Shu, Guangwen Yang, Wei He, Sibo Fang, Tianyu Guo, Kai Han, Chao Xu, Hanting Chen, Xinghao Chen, Yunhe Wang
Abstract: Diffusion Language Models (DLMs) enable fast generation, yet training large DLMs from scratch is costly. As a practical shortcut, adapting off-the-shelf Auto-Regressive (AR) model weights into a DLM could quickly equip the DLM with strong long-context generation capabilies. Prior "adaptation" attempts either modify logits or randomly grow attention masks to Full-Sequence diffusion, or simply transplant AR weights into a Block-Diffusion recipe, leaving two key questions unaddressed: where is the final destination of adaptation, and how to adapt better? For manifold benefits, we reframe the whole AR-to-DLM adaptation under the Block-Diffusion paradigm, transitioning from block size 1 to the final Block-Diffusion state. Concretely, the principled pathway of adaptation is designed as follows: we keep a context-causal path where causal attention is kept in the prefix, an efficient parallel adaptation procedure where an AR guidance is maintained, and gradual increment of the generation block size for a smoother transition. Built on these components, the adaptation is proved competitive on various models at different scales. With better adaptation, we propose NBDiff-7B that could inherit the long-context modeling and reasoning capabilities, and achieve state-of-the-art performance among the 7B-class DLMs. Codes: https://github.com/YuchuanTian/NBDiff.
Authors: Sijia Li, Yuchen Huang, Zifan Liu, Zijian Li, Jingjing fu, Lei Song, Jiang Bian, Jun Zhang, Rui Wang
Abstract: As intents unfold and environments change, multi-turn agents face continuously shifting decision contexts. Although reusing past experience is intuitively appealing, existing approaches remain limited: full trajectories are often too context-specific to transfer, while tool-level reuse ignores the surrounding context and environment. In this paper, we introduce a hybrid episodic-procedural memory strategy (H-EPM) that enables experience-induced self-evolution of multi-turn tool-use policies by adaptively reusing partially overlapping successful experiences during both inference and training. Inspired by human episodic-procedural integration, we construct a tool graph from accumulated trajectories, where recurring tool-to-tool dependencies capture procedural routines and each edge is augmented with compact episodic summaries of relevant context. At inference time, the agent dynamically balances episodic recall for contextual reasoning with procedural execution for routine steps. Beyond inference, H-EPM introduces a memory-guided reinforcement learning paradigm that directly addresses a core challenge in multi-turn agent reinforcement learning, namely ineffective exploration over long trajectories. By biasing exploration toward historically successful tool transitions, H-EPM learns a stronger policy that generalizes at inference time without relying on domain-specific experience collection. Experiments show that H-EPM consistently delivers substantial inference-time gains over strong baselines across multi-turn tool-use benchmarks, reaching improvements of up to fifty percent. It also improves reinforcement learning policy performance, achieving gains of up to forty percent on out-of-distribution tasks.
Authors: Haldun Balim, Na Li, Yilun Du
Abstract: Offline decision-making via diffusion models often produces trajectories that are misaligned with system dynamics, limiting their reliability for control. We propose Model Predictive Diffuser (MPDiffuser), a compositional diffusion framework that combines a diffusion planner with a dynamics diffusion model to generate task-aligned and dynamically plausible trajectories. MPDiffuser interleaves planner and dynamics updates during sampling, progressively correcting feasibility while preserving task intent. A lightweight ranking module then selects trajectories that best satisfy task objectives. The compositional design improves sample efficiency and adaptability by enabling the dynamics model to leverage diverse and previously unseen data independently of the planner. Empirically, we demonstrate consistent improvements over prior diffusion-based methods on unconstrained (D4RL) and constrained (DSRL) benchmarks, and validate practicality through deployment on a real quadrupedal robot.
Authors: Nicolas Tacheny
Abstract: Iterative LLM systems(self-refinement, chain-of-thought, autonomous agents) are increasingly deployed, yet their temporal dynamics remain uncharacterized. Prior work evaluates task performance at convergence but ignores the trajectory: how does semantic content evolve across iterations? Does it stabilize, drift, or oscillate? Without answering these questions, we cannot predict system behavior, guarantee stability, or systematically design iterative architectures. We formalize agentic loops as discrete dynamical systems in semantic space. Borrowing from dynamical systems theory, we define trajectories, attractors and dynamical regimes for recursive LLM transformations, providing rigorous geometric definitions adapted to this setting. Our framework reveals that agentic loops exhibit classifiable dynamics: contractive (convergence toward stable semantic attractors), oscillatory (cycling among attractors), or exploratory (unbounded divergence). Experiments on singular loops validate the framework. Iterative paraphrasing produces contractive dynamics with measurable attractor formation and decreasing dispersion. Iterative negation produces exploratory dynamics with no stable structure. Crucially, prompt design directly controls the dynamical regime - the same model exhibits fundamentally different geometric behaviors depending solely on the transformation applied. This work establishes that iterative LLM dynamics are predictable and controllable, opening new directions for stability analysis, trajectory forecasting, and principled design of composite loops that balance convergence and exploration.
Authors: Bj\"orn Deiseroth, Max Henning H\"oth, Kristian Kersting, Letitia Parcalabescu
Abstract: Retrieval-augmented generation (RAG) relies on retrieved context to guide large language models (LLM), yet treats retrieval as a weak heuristic rather than verifiable evidence -- leading to unsupported answers, hallucinations, and reliance on spurious context. We introduce a novel training framework that treats the RAG pipeline as an interactive proof system by adapting the Merlin-Arthur (M/A) protocol: Arthur (the generator LLM) trains on questions with unknown context provenance and Merlin gives helpful evidence, while Morgana injects adversarial, misleading context. Both use an XAI method to identify and modify evidence most influential to Arthur. This trains Arthur to (1) answer when evidence supports the answer, (2) reject when evidence is insufficient, and (3) rely on the context spans that truly ground the answer. We further introduce a verification framework that disentangles explanation fidelity from model predictive errors, and introduce the Explained Information Fraction (EIF), which normalizes M/A mutual-information guarantees. Across three RAG datasets and multiple LLM families and sizes, M/A training makes LLMs more grounded in evidence, increases information theoretic measures (soundness, completeness) and reject behavior with less hallucinations, without manually annotated unanswerable samples. Finally, the retriever also improves recall and MRR via automatically generated M/A hard positives and negatives. While high accuracy does not guarantee entropy flow from context to answer, our EIF results show that autonomous interactive-proof-style supervision enables RAG systems that treat retrieved documents as verifiable evidence. % rather than suggestions.
Authors: Ruifeng Tan, Weixiang Hong, Jia Li, Jiaqiang Huang, Tong-Yi Zhang
Abstract: Early prediction of battery cycle life is essential for accelerating battery research, manufacturing, and deployment. Although machine learning methods have shown encouraging results, progress is hindered by data scarcity and heterogeneity arising from diverse aging conditions. In other fields, foundation models (FMs) trained on diverse datasets have achieved broad generalization through transfer learning, but no FMs have been reported for battery cycle life prediction yet. Here we present the Pretrained Battery Transformer (PBT), the first FM for battery life prediction, developed through domain-knowledge-encoded mixture-of-expert layers. Validated on the largest public battery life database, PBT learns transferable representations from 13 lithium-ion battery (LIB) datasets, outperforming existing models by an average of 19.8%. With transfer learning, PBT achieves state-of-the-art performance across 15 diverse datasets encompassing 995 batteries and 537 aging conditions of LIBs, sodium-ion batteries and Zinc-ion batteries. This work establishes a foundation model pathway for battery lifetime prediction, paving the way toward universal battery lifetime prediction systems.
Authors: Aaron Defazio, Konstantin Mishchenko, Parameswaran Raman, Hao-Jun Michael Shi, Lin Xiao
Abstract: We propose Generalized Primal Averaging (GPA), an extension of Nesterov's method that unifies and generalizes recent averaging-based optimizers like single-worker DiLoCo and Schedule-Free, within a non-distributed setting. While DiLoCo relies on a memory-intensive two-loop structure to periodically aggregate pseudo-gradients using Nesterov momentum, GPA eliminates this complexity by decoupling Nesterov's interpolation constants to enable smooth iterate averaging at every step. Structurally, GPA resembles Schedule-Free but replaces uniform averaging with exponential moving averaging. Empirically, GPA consistently outperforms single-worker DiLoCo and AdamW with reduced memory overhead. GPA achieves speedups of 8.71%, 10.13%, and 9.58% over the AdamW baseline in terms of steps to reach target validation loss for Llama-160M, 1B, and 8B models, respectively. Similarly, on the ImageNet ViT workload, GPA achieves speedups of 7% and 25.5% in the small and large batch settings respectively. Furthermore, we prove that for any base optimizer with $O(\sqrt{T})$ regret, where $T$ is the number of iterations, GPA matches or exceeds the original convergence guarantees depending on the interpolation constants.
Authors: Qiushi Han, David Simchi-Levi, Renfei Tan, Zishuo Zhao
Abstract: We study a sequential mechanism design problem in which a principal seeks to elicit truthful reports from multiple rational agents while starting with no prior knowledge of agents' beliefs. We introduce Distributionally Robust Adaptive Mechanism (DRAM), a general framework combining insights from both mechanism design and online learning to jointly address truthfulness and cost-optimality. Throughout the sequential game, the mechanism estimates agents' beliefs and iteratively updates a distributionally robust linear program with shrinking ambiguity sets to reduce payments while preserving truthfulness. Our mechanism guarantees truthful reporting with high probability while achieving $\tilde{O}(\sqrt{T})$ cumulative regret, and we establish a matching lower bound showing that no truthful adaptive mechanism can asymptotically do better. The framework generalizes to plug-in estimators, supporting structured priors and delayed feedback. To our knowledge, this is the first adaptive mechanism under general settings that maintains truthfulness and achieves optimal regret when incentive constraints are unknown and must be learned.
Authors: Ankit Kanwar, Dominik Wagner, Luke Ong
Abstract: In safety-critical domains, reinforcement learning (RL) agents must often satisfy strict, zero-cost safety constraints while accomplishing tasks. Existing model-free methods frequently either fail to achieve near-zero safety violations or become overly conservative. We introduce Safety-Biased Trust Region Policy Optimisation (SB-TRPO), a principled algorithm for hard-constrained RL that dynamically balances cost reduction with reward improvement. At each step, SB-TRPO updates via a dynamic convex combination of the reward and cost natural policy gradients, ensuring a fixed fraction of optimal cost reduction while using remaining update capacity for reward improvement. Our method comes with formal guarantees of local progress on safety, while still improving reward whenever gradients are suitably aligned. Experiments on standard and challenging Safety Gymnasium tasks demonstrate that SB-TRPO consistently achieves the best balance of safety and task performance in the hard-constrained regime.
Authors: Yifan Zhang, Yifeng Liu, Mengdi Wang, Quanquan Gu
Abstract: The effectiveness of deep residual networks hinges on the identity shortcut connection. While this mechanism alleviates the vanishing-gradient problem, it also has a strictly additive inductive bias on feature transformations, limiting the network's ability to model complex hidden state transitions. In this paper, we introduce \textbf{Deep Delta Learning (DDL)}, which generalizes the shortcut from a fixed identity map to a learnable, state-dependent linear operator. The resulting Delta Operator is a rank-1 perturbation of the identity, $\mathbf{A}(\mathbf{X}) = \mathbf{I}- \beta(\mathbf{X})\mathbf{k} (\mathbf{X}) \mathbf{k} (\mathbf{X})^\top$, parameterized by a unit direction $\mathbf{k}(\mathbf{X})$ and a scalar gate $\beta(\mathbf{X})$. We provide a spectral analysis showing that $\beta(\mathbf{X})$ continuously interpolates the shortcut between identity ($\beta=0$), orthogonal projection ($\beta=1$), and Householder reflection ($\beta=2$). Furthermore, we rewrite the residual update as a synchronized rank-1 delta write: $\beta$ scales both the removal of the current $\mathbf{k}$-component and the injection of the new $\mathbf{k}$-component. This unification enables explicit control of the shortcut spectrum along a data-dependent direction while retaining stable training behavior. Empirically, replacing Transformer residual additions with DDL improves validation loss and perplexity, as well as downstream evaluation accuracy on language modeling tasks, with larger gains in the expanded-state setting.
Authors: Haonan Song, Qingchen Xie, Huan Zhu, Feng Xiao, Luxi Xing, Liu Kang, Fuzhen Li, Zhiyong Zheng, Feng Jiang, Ziheng Li, Kun Yan, Qingyi Si, Yanghua Xiao, Hongcheng Guo, Fan Yang
Abstract: Generative Reward Models (GRMs) have demonstrated strong performance in reward modeling, due to their interpretability and potential for refinement through reinforcement learning (RL). However, widely used pairwise GRMs create a computational bottleneck in reinforcement learning from human feedback (RLHF), when calibrating or aggregating preference signals over n candidates, often incurring O(n^2) pairwise judgments. To address this issue, we propose Intergroup Relative Preference Modeling (IRPM), an RL-based method that extends the Bradley--Terry preference-learning paradigm via intergroup comparisons to train pointwise GRMs from pairwise preference data. IRPM derives pointwise reward for each response by contrasting groups of chosen vs. rejected samples, enabling pointwise scores comparable across candidate sets and O(n) reward evaluation for a variable number of candidates during RL training, while preserving interpretability and scalability. Experiments show that IRPM achieves state-of-the-art performance among pointwise GRMs on RM-Bench, JudgeBench and RewardBench, and approaches the performance of leading pairwise GRMs. In addition, IRPM achieves substantial gains in post-training evaluations, demonstrating its effectiveness.
Authors: Maja Waldron
Abstract: One-shot prediction enables rapid adaptation of pretrained foundation models to new tasks using only one labeled example, but lacks principled uncertainty quantification. While conformal prediction provides finite-sample coverage guarantees, standard split conformal methods are inefficient in the one-shot setting due to data splitting and reliance on a single predictor. We propose Conformal Aggregation of One-Shot Predictors (CAOS), a conformal framework that adaptively aggregates multiple one-shot predictors and uses a leave-one-out calibration scheme to fully exploit scarce labeled data. Despite violating classical exchangeability assumptions, we prove that CAOS achieves valid marginal coverage using a monotonicity-based argument. Experiments on one-shot facial landmarking and RAFT text classification tasks show that CAOS produces substantially smaller prediction sets than split conformal baselines while maintaining reliable coverage.
Authors: Hengyu Shen, Tiancheng Gu, Bin Qin, Lan Wu, Yuling Wu, Shuo Tan, Zelong Sun, Jun Wang, Nan Wu, Xiang An, Weidong Cai, Ziyong Feng, Kaicheng Yang
Abstract: Vision-Language Pre-training (VLP) models have achieved remarkable success by leveraging large-scale image-text pairs. While English-centric models like CLIP and SigLIP benefit from massive datasets (e.g., LAION-400M), the development of Chinese VLP remains bottlenecked by the lack of high-quality, large-scale open-source data. In this paper, we present DanQing, a large-scale Chinese cross-modal dataset containing 100 million high-quality image-text pairs curated from Common Crawl. To ensure superior data quality, we develop an effective systematic pipeline comprising data source selection, text refinement, visual diversification, and cross-modal cross-batch filtering, thereby effectively mitigating the intrinsic noise prevalent in web data. Notably, DanQing incorporates data from 2024-2025, enabling models to capture contemporary semantic trends and emerging concepts. Extensive experiments via continued pretraining of SigLIP2 models demonstrate that DanQing consistently outperforms existing Chinese datasets across diverse downstream tasks, including zero-shot classification, cross-modal retrieval, and Chinese-centric large multimodal model tasks. Furthermore, in-depth analysis of DanQing reveals that it exhibits a more balanced semantic distribution and superior scaling capability compared to existing datasets. To facilitate further research in Chinese vision-language pre-training, we will open-source the DanQing dataset under the Creative Common CC-BY 4.0 license.
Authors: Jingkang Liang, Niklas Groll, G\"urkan Sin
Abstract: Modern process simulators enable detailed process design, simulation, and optimization; however, constructing and interpreting simulations is time-consuming and requires expert knowledge. This limits early exploration by inexperienced users. To address this, a large language model (LLM) agent is integrated with AVEVA Process Simulation (APS) via Model Context Protocol (MCP), allowing natural language interaction with rigorous process simulations. An MCP server toolset enables the LLM to communicate programmatically with APS using Python, allowing it to execute complex simulation tasks from plain-language instructions. Two water-methanol separation case studies assess the framework across different task complexities and interaction modes. The first shows the agent autonomously analyzing flowsheets, finding improvement opportunities, and iteratively optimizing, extracting data, and presenting results clearly. The framework benefits both educational purposes, by translating technical concepts and demonstrating workflows, and experienced practitioners by automating data extraction, speeding routine tasks, and supporting brainstorming. The second case study assesses autonomous flowsheet synthesis through both a step-by-step dialogue and a single prompt, demonstrating its potential for novices and experts alike. The step-by-step mode gives reliable, guided construction suitable for educational contexts; the single-prompt mode constructs fast baseline flowsheets for later refinement. While current limitations such as oversimplification, calculation errors, and technical hiccups mean expert oversight is still needed, the framework's capabilities in analysis, optimization, and guided construction suggest LLM-based agents can become valuable collaborators.
Authors: Yifei Zhang, Hooshang Nayyeri, Rinat Khaziev, Emine Yilmaz, Gokhan Tur, Dilek Hakkani-T\"ur, Hari Thadakamalla
Abstract: Recent advances in task-oriented dialogue (TOD) systems, driven by large language models (LLMs) with extensive API and tool integration, have enabled conversational agents to coordinate interleaved goals, maintain long-horizon context, and act proactively through asynchronous execution. These capabilities extend beyond traditional TOD systems, yet existing benchmarks lack systematic support for evaluating such agentic behaviors. To address this gap, we introduce ATOD, a benchmark and synthetic dialogue generation pipeline that produces richly annotated conversations requiring long-term reasoning. ATOD captures key characteristics of advanced TOD, including multi-goal coordination, dependency management, memory, adaptability, and proactivity. Building on ATOD, we propose ATOD-Eval, a holistic evaluation framework that translates these dimensions into fine-grained metrics and supports reproducible offline and online evaluation. We further present a strong agentic memory-based evaluator for benchmarking on ATOD. Experiments show that ATOD-Eval enables comprehensive assessment across task completion, agentic capability, and response quality, and that the proposed evaluator offers a better accuracy-efficiency tradeoff compared to existing memory- and LLM-based approaches under this evaluation setting.
Authors: Shreya Rajpal, Michal Golovanevsky, Carsten Eickhoff
Abstract: Video summarization helps turn long videos into clear, concise representations that are easier to review, document, and analyze, especially in high-stakes domains like surgical training. Prior work has progressed from using basic visual features like color, motion, and structural changes to using pre-trained vision-language models that can better understand what's happening in the video (semantics) and capture temporal flow, resulting in more context-aware video summarization. We propose a three-stage framework, PRISM: Procedural Representation via Integrated Semantic and Multimodal analysis, that produces semantically grounded video summaries. PRISM combines adaptive visual sampling, label-driven keyframe anchoring, and contextual validation using a large language model (LLM). Our method ensures that selected frames reflect meaningful and procedural transitions while filtering out generic or hallucinated content, resulting in contextually coherent summaries across both domain-specific and instructional videos. We evaluate our method on instructional and activity datasets, using reference summaries for instructional videos. Despite sampling fewer than 5% of the original frames, our summaries retain 84% semantic content while improving over baselines by as much as 33%. Our approach generalizes across procedural and domain-specific video tasks, achieving strong performance with both semantic alignment and precision.
Authors: Aravind B, Anirud R. S., Sai Surya Teja N, Bala Subrahmanya Sriranga Navaneeth A, Karthika R, Mohankumar N
Abstract: Class imbalance refers to a situation where certain classes in a dataset have significantly fewer samples than oth- ers, leading to biased model performance. Class imbalance in network intrusion detection using Tabular Denoising Diffusion Probability Models (TabDDPM) for data augmentation is ad- dressed in this paper. Our approach synthesizes high-fidelity minority-class samples from the CIC-IDS2017 dataset through iterative denoising processes. For the minority classes that have smaller samples, synthetic samples were generated and merged with the original dataset. The augmented training data enables an ANN classifier to achieve near-perfect recall on previously underrepresented attack classes. These results establish diffusion models as an effective solution for tabular data imbalance in security domains, with potential applications in fraud detection and medical diagnostics.
Authors: Xu Zhang, Junwei Deng, Chang Xu, Hao Li, Jiang Bian
Abstract: Time series generation (TSG) is widely used across domains, yet most existing methods assume regular sampling and fixed output resolutions. These assumptions are often violated in practice, where observations are irregular and sparse, while downstream applications require continuous and high-resolution TS. Although Neural Controlled Differential Equation (NCDE) is promising for modeling irregular TS, it is constrained by a single dynamics function, tightly coupled optimization, and limited ability to adapt learned dynamics to newly generated samples from the generative model. We propose Diff-MN, a continuous TSG framework that enhances NCDE with a Mixture-of-Experts (MoE) dynamics function and a decoupled architectural design for dynamics-focused training. To further enable NCDE to generalize to newly generated samples, Diff-MN employs a diffusion model to parameterize the NCDE temporal dynamics parameters (MoE weights), i.e., jointly learn the distribution of TS data and MoE weights. This design allows sample-specific NCDE parameters to be generated for continuous TS generation. Experiments on ten public and synthetic datasets demonstrate that Diff-MN consistently outperforms strong baselines on both irregular-to-regular and irregular-to-continuous TSG tasks. The code is available at the link https://github.com/microsoft/TimeCraft/tree/main/Diff-MN.
URLs: https://github.com/microsoft/TimeCraft/tree/main/Diff-MN.
Authors: Fan Huang, Haewoon Kwak, Jisun An
Abstract: Large Language Models (LLMs) are increasingly employed in various question-answering tasks. However, recent studies showcase that LLMs are susceptible to persuasion and could adopt counterfactual beliefs. We present a systematic evaluation of LLM susceptibility to persuasion under the Source--Message--Channel--Receiver (SMCR) communication framework. Across five mainstream Large Language Models (LLMs) and three domains (factual knowledge, medical QA, and social bias), we analyze how different persuasive strategies influence belief stability over multiple interaction turns. We further examine whether meta-cognition prompting (i.e., eliciting self-reported confidence) affects resistance to persuasion. Results show that the smallest model (Llama 3.2-3B) exhibits extreme compliance, with 82.5% of belief changes occurring at the first persuasive turn (average end turn of 1.1--1.4). Contrary to expectations, meta-cognition prompting increases vulnerability by accelerating belief erosion rather than enhancing robustness. Finally, we evaluate adversarial fine-tuning as a defense. While GPT-4o-mini achieves near-complete robustness (98.6%) and Mistral~7B improves substantially (35.7% $\rightarrow$ 79.3%), Llama models remain highly susceptible (<14%) even when fine-tuned on their own failure cases. Together, these findings highlight substantial model-dependent limits of current robustness interventions and offer guidance for developing more trustworthy LLMs.
Authors: Hsin-Yi Lin, Huan-Hsin Tseng, Samuel Yen-Chi Chen, Shinjae Yoo
Abstract: Super-resolution (SR) seeks to reconstruct high-resolution (HR) data from low-resolution (LR) observations. Classical deep learning methods have advanced SR substantially, but require increasingly deeper networks, large datasets, and heavy computation to capture fine-grained correlations. In this work, we present the \emph{first study} to investigate quantum circuits for SR. We propose a framework based on Variational Quantum Circuits (VQCs) with \emph{Adaptive Non-Local Observable} (ANO) measurements. Unlike conventional VQCs with fixed Pauli readouts, ANO introduces trainable multi-qubit Hermitian observables, allowing the measurement process to adapt during training. This design leverages the high-dimensional Hilbert space of quantum systems and the representational structure provided by entanglement and superposition. Experiments demonstrate that ANO-VQCs achieve up to five-fold higher resolution with a relatively small model size, suggesting a promising new direction at the intersection of quantum machine learning and super-resolution.
Authors: Peiran Li, Fangzhou Lin, Shuo Xing, Xiang Zheng, Xi Hong, Siyuan Yang, Jiashuo Sun, Zhengzhong Tu, Chaoqun Ni
Abstract: Citations are the bedrock of scientific authority, yet their integrity is compromised by widespread miscitations: ranging from nuanced distortions to fabricated references. Systematic citation verification is currently unfeasible; manual review cannot scale to modern publishing volumes, while existing automated tools are restricted by abstract-only analysis or small-scale, domain-specific datasets in part due to the "paywall barrier" of full-text access. We introduce BibAgent, a scalable, end-to-end agentic framework for automated citation verification. BibAgent integrates retrieval, reasoning, and adaptive evidence aggregation, applying distinct strategies for accessible and paywalled sources. For paywalled references, it leverages a novel Evidence Committee mechanism that infers citation validity via downstream citation consensus. To support systematic evaluation, we contribute a 5-category Miscitation Taxonomy and MisciteBench, a massive cross-disciplinary benchmark comprising 6,350 miscitation samples spanning 254 fields. Our results demonstrate that BibAgent outperforms state-of-the-art Large Language Model (LLM) baselines in citation verification accuracy and interpretability, providing scalable, transparent detection of citation misalignments across the scientific literature.
Authors: Hao Li, He Cao, Shenyao Peng, Zijing Liu, Bin Feng, Yu Wang, Zhiyuan Yan, Yonghong Tian, Yu Li, Li Yuan
Abstract: Language models are revolutionizing the biochemistry domain, assisting scientists in drug design and chemical synthesis with high efficiency. Yet current approaches struggle between small language models prone to hallucination and limited knowledge retention, and large cloud-based language models plagued by privacy risks and high inference costs. To bridge this gap, we introduce ChemCRAFT, a novel framework leveraging agentic reinforcement learning to decouple chemical reasoning from knowledge storage. Instead of forcing the model to memorize vast chemical data, our approach empowers the language model to interact with a sandbox for precise information retrieval. This externalization of knowledge allows a locally deployable small model to achieve superior performance with minimal inference costs. To enable small language models for agent-calling ability, we build an agentic trajectory construction pipeline and a comprehensive chemical-agent sandbox. Based on sandbox interactions, we constructed ChemToolDataset, the first large-scale chemical tool trajectory dataset. Simultaneously, we propose SMILES-GRPO to build a dense chemical reward function, promoting the model's ability to call chemical agents. Evaluations across diverse aspects of drug design show that ChemCRAFT outperforms current cloud-based LLMs in molecular structure analysis, molecular optimization, and synthesis pathway prediction, demonstrating that scientific reasoning is not solely an emergent ability of model scale, but a learnable policy of tool orchestration. This work establishes a cost-effective and privacy-preserving paradigm for AI-aided chemistry, opening new avenues for accelerating molecular discovery with locally deployable agents. Code available at https://github.com/HowardLi1984/ChemCraft.
Authors: Raja Gond, Aditya K Kamath, Ramachandran Ramjee, Ashish Panwar
Abstract: In LLM inference, the same prompt may yield different outputs across different runs. At the system level, this non-determinism arises from floating-point non-associativity combined with dynamic batching and GPU kernels whose reduction orders vary with batch size. A straightforward way to eliminate non-determinism is to disable dynamic batching during inference, but doing so severely degrades throughput. Another approach is to make kernels batch-invariant; however, this tightly couples determinism to kernel design, requiring new implementations. This coupling also imposes fixed runtime overheads, regardless of how much of the workload actually requires determinism. Inspired by ideas from speculative decoding, we present LLM-42, a scheduling-based approach to enable determinism in LLM inference. Our key observation is that if a sequence is in a consistent state, the next emitted token is likely to be consistent even with dynamic batching. Moreover, most GPU kernels use shape-consistent reductions. Leveraging these insights, LLM-42 decodes tokens using a non-deterministic fast path and enforces determinism via a lightweight verify-rollback loop. The verifier replays candidate tokens under a fixed-shape reduction schedule, commits those that are guaranteed to be consistent across runs, and rolls back those violating determinism. LLM-42 mostly re-uses existing kernels unchanged and incurs overhead only in proportion to the traffic that requires determinism.
Authors: Dezhang Kong, Zhuxi Wu, Shiqi Liu, Zhicheng Tan, Kuichen Lu, Minghao Li, Qichen Liu, Shengyu Chu, Zhenhua Xu, Xuan Liu, Meng Han
Abstract: LLM-based web agents have become increasingly popular for their utility in daily life and work. However, they exhibit critical vulnerabilities when processing malicious URLs: accepting a disguised malicious URL enables subsequent access to unsafe webpages, which can cause severe damage to service providers and users. Despite this risk, no benchmark currently targets this emerging threat. To address this gap, we propose MalURLBench, the first benchmark for evaluating LLMs' vulnerabilities to malicious URLs. MalURLBench contains 61,845 attack instances spanning 10 real-world scenarios and 7 categories of real malicious websites. Experiments with 12 popular LLMs reveal that existing models struggle to detect elaborately disguised malicious URLs. We further identify and analyze key factors that impact attack success rates and propose URLGuard, a lightweight defense module. We believe this work will provide a foundational resource for advancing the security of web agents. Our code is available at https://github.com/JiangYingEr/MalURLBench.
Authors: Zhewen Tan, Wenhan Yu, Jianfeng Si, Tongxin Liu, Kaiqi Guan, Huiyan Jin, Jiawen Tao, Xiaokun Yuan, Duohe Ma, Xiangzheng Zhang, Tong Yang, Lin Sun
Abstract: In recent years, safety risks associated with large language models have become increasingly prominent, highlighting the urgent need to mitigate the generation of toxic and harmful content. The mainstream paradigm for LLM safety alignment typically adopts a collaborative framework involving three roles: an attacker for adversarial prompt generation, a defender for safety defense, and an evaluator for response assessment. In this paper, we propose a closed-loop reinforcement learning framework called TriPlay-RL that enables iterative and co-improving collaboration among three roles with near-zero manual annotation. Experimental results show that the attacker preserves high output diversity while achieving a 20%-50% improvement in adversarial effectiveness; the defender attains 10%-30% gains in safety performance without degrading general reasoning capability; and the evaluator continuously refines its fine-grained judgment ability through iterations, accurately distinguishing unsafe responses, simple refusals, and useful guidance. Overall, our framework establishes an efficient and scalable paradigm for LLM safety alignment, enabling continuous co-evolution within a unified learning loop.
Authors: Lei Zhang, Yongda Yu, Minghui Yu, Xinxin Guo, Zhengqi Zhuang, Guoping Rong, Dong Shao, Haifeng Shen, Hongyu Kuang, Zhengfeng Li, Boge Wang, Guoan Zhang, Bangyu Xiang, Xiaobin Xu
Abstract: High-quality evaluation benchmarks are pivotal for deploying Large Language Models (LLMs) in Automated Code Review (ACR). However, existing benchmarks suffer from two critical limitations: first, the lack of multi-language support in repository-level contexts, which restricts the generalizability of evaluation results; second, the reliance on noisy, incomplete ground truth derived from raw Pull Request (PR) comments, which constrains the scope of issue detection. To address these challenges, we introduce AACR-Bench a comprehensive benchmark that provides full cross-file context across multiple programming languages. Unlike traditional datasets, AACR-Bench employs an "AI-assisted, Expert-verified" annotation pipeline to uncover latent defects often overlooked in original PRs, resulting in a 285% increase in defect coverage. Extensive evaluations of mainstream LLMs on AACR-Bench reveal that previous assessments may have either misjudged or only partially captured model capabilities due to data limitations. Our work establishes a more rigorous standard for ACR evaluation and offers new insights on LLM based ACR, i.e., the granularity/level of context and the choice of retrieval methods significantly impact ACR performance, and this influence varies depending on the LLM, programming language, and the LLM usage paradigm e.g., whether an Agent architecture is employed. The code, data, and other artifacts of our evaluation set are available at https://github.com/alibaba/aacr-bench .
Authors: Jiajie Su, Haoyuan Wang, Xiaohua Feng, Yunshan Ma, Xiaobo Xia, Yuyuan Li, Xiaolin Zheng, Jianmao Xiao, Chaochao Chen
Abstract: Knowledge editing emerges as a crucial technique for efficiently correcting incorrect or outdated knowledge in large language models (LLM). Existing editing methods rely on a rigid mapping from parameter or module modifications to output, which causes the generalization limitation in Multimodal LLM (MLLM). In this paper, we reformulate MLLM editing as an out-of-distribution (OOD) generalization problem, where the goal is to discern semantic shift with factual shift and thus achieve robust editing among diverse cross-modal prompting. The key challenge of this OOD problem lies in identifying invariant causal trajectories that generalize accurately while suppressing spurious correlations. To address it, we propose ODEdit, a plug-and-play invariant learning based framework that optimizes the tripartite OOD risk objective to simultaneously enhance editing reliability, locality, and generality.We further introduce an edit trajectory invariant learning method, which integrates a total variation penalty into the risk minimization objective to stabilize edit trajectories against environmental variations. Theoretical analysis and extensive experiments demonstrate the effectiveness of ODEdit.
Authors: M. Naser Lessani, Zhenlong Li, Manzhu Yu, Helen Greatrex, Chan Shen
Abstract: The first law of geography is a cornerstone of spatial analysis, emphasizing that nearby and related locations tend to be more similar, however, defining what constitutes "near" and "related" remains challenging, as different phenomena exhibit distinct spatial patterns. Traditional local regression models, such as Geographically Weighted Regression (GWR) and Multiscale GWR (MGWR), quantify spatial relationships solely through geographic proximity. In an era of globalization and digital connectivity, however, geographic proximity alone may be insufficient to capture how locations are interconnected. To address this limitation, we propose a new multiscale local regression framework, termed M-SGWR, which characterizes spatial interaction across two dimensions: geographic proximity and attribute (variable) similarity. For each predictor, geographic and attribute-based weight matrices are constructed separately and then combined using an optimized parameter, alpha, which governs their relative contribution to local model fitting. Analogous to variable-specific bandwidths in MGWR, the optimal alpha varies by predictor, allowing the model to flexibly account for geographic, mixed, or non-spatial (remote similarity) effects. Results from two simulation experiments and one empirical application demonstrate that M-SGWR consistently outperforms GWR, SGWR, and MGWR across all goodness-of-fit metrics.
Authors: Shih-Hsuan Chiu, Ming-Syan Chen
Abstract: Personalized virtual assistants powered by large language models (LLMs) on edge devices are attracting growing attention, with Retrieval-Augmented Generation (RAG) emerging as a key method for personalization by retrieving relevant profile data and generating tailored responses. However, deploying RAG on edge devices faces efficiency hurdles due to the rapid growth of profile data, such as user-LLM interactions and recent updates. While Computing-in-Memory (CiM) architectures mitigate this bottleneck by eliminating data movement between memory and processing units via in-situ operations, they are susceptible to environmental noise that can degrade retrieval precision. This poses a critical issue in dynamic, multi-domain edge-based scenarios (e.g., travel, medicine, and law) where both accuracy and adaptability are paramount. To address these challenges, we propose Task-Oriented Noise-resilient Embedding Learning (TONEL), a framework that improves noise robustness and domain adaptability for RAG in noisy edge environments. TONEL employs a noise-aware projection model to learn task-specific embeddings compatible with CiM hardware constraints, enabling accurate retrieval under noisy conditions. Extensive experiments conducted on personalization benchmarks demonstrate the effectiveness and practicality of our methods relative to strong baselines, especially in task-specific noisy scenarios.
Authors: Emanuele Del Sozzo, Martin Fleming, Kenneth Flamm, Neil Thompson
Abstract: Graphics Processing Units (GPUs) are the state-of-the-art architecture for essential tasks, ranging from rendering 2D/3D graphics to accelerating workloads in supercomputing centers and, of course, Artificial Intelligence (AI). As GPUs continue improving to satisfy ever-increasing performance demands, analyzing past and current progress becomes paramount in determining future constraints on scientific research. This is particularly compelling in the AI domain, where rapid technological advancements and fierce global competition have led the United States to recently implement export control regulations limiting international access to advanced AI chips. For this reason, this paper studies technical progress in NVIDIA datacenter GPUs released from the mid-2000s until today. Specifically, we compile a comprehensive dataset of datacenter NVIDIA GPUs comprising several features, ranging from computational performance to release price. Then, we examine trends in main GPU features and estimate progress indicators for per-memory bandwidth, per-dollar, and per-watt increase rates. Our main results identify doubling times of 1.44 and 1.69 years for FP16 and FP32 operations (without accounting for sparsity benefits), while FP64 doubling times range from 2.06 to 3.79 years. Off-chip memory size and bandwidth grew at slower rates than computing performance, doubling every 3.32 to 3.53 years. The release prices of datacenter GPUs have roughly doubled every 5.1 years, while their power consumption has approximately doubled every 16 years. Finally, we quantify the potential implications of current U.S. export control regulations in terms of the potential performance gaps that would result if implementation were assumed to be complete and successful. We find that recently proposed changes to export controls would shrink the potential performance gap from 23.6x to 3.54x.
Authors: Dimitris Michailidis, Sennay Ghebreab, Fernando P. Santos
Abstract: Who benefits from expanding transport networks? While designed to improve mobility, such interventions can also create inequality. In this paper, we show that disparities arise not only from the structure of the network itself but also from differences in how commuters adapt to it. We model commuters as reinforcement learning agents who adapt their travel choices at different learning rates, reflecting unequal access to resources and information. To capture potential efficiency-fairness tradeoffs, we introduce the Price of Learning (PoL), a measure of inefficiency during learning. We analyze both a stylized network -- inspired in the well-known Braess's paradox, yet with two-source nodes -- and an abstraction of a real-world metro system (Amsterdam). Our simulations show that network expansions can simultaneously increase efficiency and amplify inequality, especially when faster learners disproportionately benefit from new routes before others adapt. These results highlight that transport policies must account not only for equilibrium outcomes but also for the heterogeneous ways commuters adapt, since both shape the balance between efficiency and fairness.
Authors: Steve Azzolin, Stefano Teso, Bruno Lepri, Andrea Passerini, Sagar Malhotra
Abstract: Explanations provided by Self-explainable Graph Neural Networks (SE-GNNs) are fundamental for understanding the model's inner workings and for identifying potential misuse of sensitive attributes. Although recent works have highlighted that these explanations can be suboptimal and potentially misleading, a characterization of their failure cases is unavailable. In this work, we identify a critical failure of SE-GNN explanations: explanations can be unambiguously unrelated to how the SE-GNNs infer labels. We show that, on the one hand, many SE-GNNs can achieve optimal true risk while producing these degenerate explanations, and on the other, most faithfulness metrics can fail to identify these failure modes. Our empirical analysis reveals that degenerate explanations can be maliciously planted (allowing an attacker to hide the use of sensitive attributes) and can also emerge naturally, highlighting the need for reliable auditing. To address this, we introduce a novel faithfulness metric that reliably marks degenerate explanations as unfaithful, in both malicious and natural settings. Our code is available in the supplemental.
Authors: Jie Liu, Yu Sun, Alpar Cseke, Yao Feng, Nicolas Heron, Michael J. Black, Yan Zhang
Abstract: Generating 3D humans that functionally interact with 3D scenes remains an open problem with applications in embodied AI, robotics, and interactive content creation. The key challenge involves reasoning about both the semantics of functional elements in 3D scenes and the 3D human poses required to achieve functionality-aware interaction. Unfortunately, existing methods typically lack explicit reasoning over object functionality and the corresponding human-scene contact, resulting in implausible or functionally incorrect interactions. In this work, we propose FunHSI, a training-free, functionality-driven framework that enables functionally correct human-scene interactions from open-vocabulary task prompts. Given a task prompt, FunHSI performs functionality-aware contact reasoning to identify functional scene elements, reconstruct their 3D geometry, and model high-level interactions via a contact graph. We then leverage vision-language models to synthesize a human performing the task in the image and estimate proposed 3D body and hand poses. Finally, the proposed 3D body configuration is refined via stage-wise optimization to ensure physical plausibility and functional correctness. In contrast to existing methods, FunHSI not only synthesizes more plausible general 3D interactions, such as "sitting on a sofa'', while supporting fine-grained functional human-scene interactions, e.g., "increasing the room temperature''. Extensive experiments demonstrate that FunHSI consistently generates functionally correct and physically plausible human-scene interactions across diverse indoor and outdoor scenes.
Authors: Zhengzheng Tang
Abstract: Spiking Neural Networks (SNNs) promise energy-efficient computing through event-driven sparsity, yet all existing approaches sacrifice accuracy by approximating continuous values with discrete spikes. We propose NEXUS, a framework that achieves bit-exact ANN-to-SNN equivalence -- not approximate, but mathematically identical outputs. Our key insight is constructing all arithmetic operations, both linear and nonlinear, from pure IF neuron logic gates that implement IEEE-754 compliant floating-point arithmetic. Through spatial bit encoding (zero encoding error by construction), hierarchical neuromorphic gate circuits (from basic logic gates to complete transformer layers), and surrogate-free STE training (exact identity mapping rather than heuristic approximation), NEXUS produces outputs identical to standard ANNs up to machine precision. Experiments on models up to LLaMA-2 70B demonstrate identical task accuracy (0.00% degradation) with mean ULP error of only 6.19, while achieving 27-168,000$\times$ energy reduction on neuromorphic hardware. Crucially, spatial bit encoding's single-timestep design renders the framework inherently immune to membrane potential leakage (100% accuracy across all decay factors $\beta\in[0.1,1.0]$), while tolerating synaptic noise up to $\sigma=0.2$ with >98% gate-level accuracy.
Authors: Albert Tseng, Christopher De Sa
Abstract: Modern sparse language models typically achieve sparsity through Mixture-of-Experts (MoE) layers, which dynamically route tokens to dense MLP "experts." However, dynamic hard routing has a number of drawbacks, such as potentially poor hardware efficiency and needing auxiliary losses for stable training. In contrast, the tokenizer embedding table, which is natively sparse, largely avoids these issues by selecting a single embedding per token at the cost of not having contextual information. In this work, we introduce the Large Lookup Layer (L$^3$), which unlocks a new axis of sparsity by generalizing embedding tables to model decoder layers. L$^3$ layers use static token-based routing to aggregate a set of learned embeddings per token in a context-dependent way, allowing the model to efficiently balance memory and compute by caching information in embeddings. L$^3$ has two main components: (1) a systems-friendly architecture that allows for fast training and CPU-offloaded inference with no overhead, and (2) an information-theoretic embedding allocation algorithm that effectively balances speed and quality. We empirically test L$^3$ by training transformers with up to 2.6B active parameters and find that L$^3$ strongly outperforms both dense models and iso-sparse MoEs in both language modeling and downstream tasks.
Authors: Hongxu Chen, Hongxiang Li, Zhen Wang, Long Chen
Abstract: Flow Matching (FM) models have emerged as a leading paradigm for high-fidelity synthesis. However, their reliance on iterative Ordinary Differential Equation (ODE) solving creates a significant latency bottleneck. Existing solutions face a dichotomy: training-free solvers suffer from significant performance degradation at low Neural Function Evaluations (NFEs), while training-based one- or few-steps generation methods incur prohibitive training costs and lack plug-and-play versatility. To bridge this gap, we propose the Bi-Anchor Interpolation Solver (BA-solver). BA-solver retains the versatility of standard training-free solvers while achieving significant acceleration by introducing a lightweight SideNet (1-2% backbone size) alongside the frozen backbone. Specifically, our method is founded on two synergistic components: \textbf{1) Bidirectional Temporal Perception}, where the SideNet learns to approximate both future and historical velocities without retraining the heavy backbone; and 2) Bi-Anchor Velocity Integration, which utilizes the SideNet with two anchor velocities to efficiently approximate intermediate velocities for batched high-order integration. By utilizing the backbone to establish high-precision ``anchors'' and the SideNet to densify the trajectory, BA-solver enables large interval sizes with minimized error. Empirical results on ImageNet-256^2 demonstrate that BA-solver achieves generation quality comparable to 100+ NFEs Euler solver in just 10 NFEs and maintains high fidelity in as few as 5 NFEs, incurring negligible training costs. Furthermore, BA-solver ensures seamless integration with existing generative pipelines, facilitating downstream tasks such as image editing.
Authors: Wonduk Seo, Wonseok Choi, Junseo Koh, Juhyeon Lee, Hyunjin An, Minhyeong Yu, Jian Park, Qingshan Zhou, Seunghyun Lee, Yi Bu
Abstract: Large Language Models (LLMs) increasingly support culturally sensitive decision making, yet often exhibit misalignment due to skewed pretraining data and the absence of structured value representations. Existing methods can steer outputs, but often lack demographic grounding and treat values as independent, unstructured signals, reducing consistency and interpretability. We propose OG-MAR, an Ontology-Guided Multi-Agent Reasoning framework. OG-MAR summarizes respondent-specific values from the World Values Survey (WVS) and constructs a global cultural ontology by eliciting relations over a fixed taxonomy via competency questions. At inference time, it retrieves ontology-consistent relations and demographically similar profiles to instantiate multiple value-persona agents, whose outputs are synthesized by a judgment agent that enforces ontology consistency and demographic proximity. Experiments on regional social-survey benchmarks across four LLM backbones show that OG-MAR improves cultural alignment and robustness over competitive baselines, while producing more transparent reasoning traces.
Authors: Amit Dhurandhar, Vijil Chenthamarakshan, Dennis Wei, Tejaswini Pedapati, Karthikeyan Natesan Ramamurthy, Rahul Nair
Abstract: Transformers are arguably the preferred architecture for language generation. In this paper, inspired by continued fractions, we introduce a new function class for generative modeling. The architecture family implementing this function class is named CoFrGeNets - Continued Fraction Generative Networks. We design novel architectural components based on this function class that can replace Multi-head Attention and Feed-Forward Networks in Transformer blocks while requiring much fewer parameters. We derive custom gradient formulations to optimize the proposed components more accurately and efficiently than using standard PyTorch-based gradients. Our components are a plug-in replacement requiring little change in training or inference procedures that have already been put in place for Transformer-based models thus making our approach easy to incorporate in large industrial workflows. We experiment on two very different transformer architectures GPT2-xl (1.5B) and Llama3 (3.2B), where the former we pre-train on OpenWebText and GneissWeb, while the latter we pre-train on the docling data mix which consists of nine different datasets. Results show that the performance on downstream classification, Q\& A, reasoning and text understanding tasks of our models is competitive and sometimes even superior to the original models with $\frac{2}{3}$ to $\frac{1}{2}$ the parameters and shorter pre-training time. We believe that future implementations customized to hardware will further bring out the true potential of our architectures.
Authors: Akash Dhasade, Anne-Marie Kermarrec, Igor Pavlovic, Diana Petrescu, Rafael Pires, Mathis Randl, Martijn de Vos
Abstract: Low-rank adaptation (LoRA) enables parameter efficient specialization of large language models (LLMs) through modular adapters, resulting in rapidly growing public adapter pools spanning diverse tasks. Effectively using these adapters requires routing: selecting and composing the appropriate adapters for a query. We introduce LORAUTER, a novel routing framework that selects and composes LoRA adapters using task representations rather than adapter characteristics. Unlike existing approaches that map queries directly to adapters, LORAUTER routes queries via task embeddings derived from small validation sets and does not require adapter training data. By operating at the task level, LORAUTER achieves efficient routing that scales with the number of tasks rather than the number of adapters. Experiments across multiple tasks show that LORAUTER consistently outperforms baseline routing approaches, matching Oracle performance (101.2%) when task-aligned adapters exist and achieving state-of-the-art results on unseen tasks (+5.2 points). We further demonstrate the robustness of LORAUTER to very large, noisy adapter pools by scaling it to over 1500 adapters.
Authors: Seongchan Park, Jaehong Kim, Hyeonseung Kim, Heejin Bin, Sue Moon, Wonjae Lee
Abstract: Understanding how media rhetoric shapes audience engagement is crucial in the attention economy. This study examines how moral emotional framing by mainstream news channels on YouTube influences user behavior across Korea and the United States. To capture the platform's multimodal nature, combining thumbnail images and video titles, we develop a multimodal moral emotion classifier by fine tuning a vision language model. The model is trained on human annotated multimodal datasets in both languages and applied to approximately 400,000 videos from major news outlets. We analyze engagement levels including views, likes, and comments, representing increasing degrees of commitment. The results show that other condemning rhetoric expressions of moral outrage that criticize others morally consistently increase all forms of engagement across cultures, with effects ranging from passive viewing to active commenting. These findings suggest that moral outrage is a particularly effective emotional strategy, attracting not only attention but also active participation. We discuss concerns about the potential misuse of other condemning rhetoric, as such practices may deepen polarization by reinforcing in group and out group divisions. To facilitate future research and ensure reproducibility, we publicly release our Korean and English multimodal moral emotion classifiers.
Authors: Yifan Zhu, Huiqiang Rong, Haoran Luo
Abstract: Large Language Models (LLMs) often hallucinate, generating content inconsistent with the input. Retrieval-Augmented Generation (RAG) and Reinforcement Learning with Human Feedback (RLHF) can mitigate hallucinations but require resource-intensive retrieval or large-scale fine-tuning. Decoding-based methods are lighter yet lack explicit hallucination control. To address this, we present Token-Guard, a token-level hallucination control method based on self-checking decoding. Token-Guard performs internal verification at each reasoning step to detect hallucinated tokens before they propagate. Candidate fragments are further evaluated in a latent space with explicit hallucination risk scoring, while iterative pruning and regeneration dynamically correct detected errors. Experiments on HALU datasets show Token-Guard substantially reduces hallucinations and improves generation accuracy, offering a scalable, modular solution for reliable LLM outputs. Our code is publicly available.