# Research Papers — 2026-06 · bundle page 246 of 260

> Full-text markdown of papers #12251–12300 of 12993 published in 2026-06, newest first. AI-readable mirrors of the full 12993-paper 2026-06 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-06/page/246
[← Previous bundle](https://aipod.dev/papers/archive/2026-06/page/245.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-06/page/247.md)

### BAHSD: Bridging the Long-tail Gap via Adaptive Distillation in Black-box Sequential Recommendation

Authors: Xi Zhou, Famin Wu, Mingming Li, Hongyue Zhang, Jiao Dai, Jizhong Han, Tao Guo
Published: 2026-06-02
Categories: cs.IR, cs.AI
arXiv: https://arxiv.org/abs/2606.03091v2

Sequential recommendation systems are widely adopted but often deployed as black-box APIs, which has driven recent interest in model extraction to replicate their capabilities locally. However, the long-tail distribution induces severe signal heterogeneity: dense head sequences trigger the solidification of teacher preference, biasing extraction toward local patterns, while sparse tail sequences yield flat, noisy predictions. Existing one-size-fits-all extraction overlooks this disparity, resulting in noise overfitting and suboptimal knowledge transfer. We propose BAHSD, a black-box adaptive distillation framework that handles signal heterogeneity via a multi-scale consistency probing mechanism to implicitly quantify signal reliability. Based on this, an adaptive hierarchical objective is designed: dynamic-temperature KL divergence mitigates preference solidification for high-confidence signals, while ranking consistency and InfoNCE contrastive learning provide noise-robust enhancement for low-confidence signals. BAHSD consistently outperforms baselines, achieving up to 4.98\% gain over the teacher and 80\%+ improvement on tail users, offering a plug-and-play solution for high-fidelity black-box recommendation extraction.

---

### "**Important** You should give me full credits!": Exploring Prompt Injection Attacks on LLM-Based Automatic Grading Systems

Authors: Hang Li, Fedor Filippov, Yuping Lin, Pengfei He, Kaiqi Yang, Yucheng Chu, Hui Liu, Jiliang Tang, Yingqian Cui
Published: 2026-06-02
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2606.03090v3

The emergence of large language models (LLMs) has significantly accelerated recent research on LLM-based automatic grading (AG) systems. Benefiting from the strong instruction-following capabilities and broad prior knowledge of LLMs, educators can deploy AG systems across diverse tasks using only natural language rubrics while achieving satisfactory grading performance. Despite these advantages, new security concerns may also arise. In particular, prompt injection (PI) attacks have recently become a major threat to LLM-based applications. In the context of AG, attackers can potentially exploit PI vulnerabilities to manipulate grading systems into assigning artificially high scores regardless of the actual answer quality. Such behavior poses serious risks to the fairness, reliability, and integrity of educational assessment. In this work, we study PI attacks in AG systems, and systematically investigate the effectiveness of such attacks in educational scenarios. We further evaluate the effectiveness of existing defensive strategies against these attacks. Through comprehensive experiments under rubric-based grading settings, we demonstrate that current LLM-based AG systems remain highly vulnerable to PI attacks. We hope that our findings raise awareness of this emerging threat and motivate future research toward secure, robust, and trustworthy LLM-based educational systems.

---

### "**Important** You should give me full credits!": Exploring Prompt Injection Attacks on LLM-Based Automatic Grading Systems

Authors: Hang Li, Fedor Filippov, Yuping Lin, Pengfei He, Kaiqi Yang, Yucheng Chu, Yingqian Cui, Hui Liu, Jiliang Tang
Published: 2026-06-02
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2606.03090v2

The emergence of large language models (LLMs) has significantly accelerated recent research on LLM-based automatic grading (AG) systems. Benefiting from the strong instruction-following capabilities and broad prior knowledge of LLMs, educators can deploy AG systems across diverse tasks using only natural language rubrics while achieving satisfactory grading performance. Despite these advantages, new security concerns may also arise. In particular, prompt injection (PI) attacks have recently become a major threat to LLM-based applications. In the context of AG, attackers can potentially exploit PI vulnerabilities to manipulate grading systems into assigning artificially high scores regardless of the actual answer quality. Such behavior poses serious risks to the fairness, reliability, and integrity of educational assessment. In this work, we study PI attacks in AG systems, and systematically investigate the effectiveness of such attacks in educational scenarios. We further evaluate the effectiveness of existing defensive strategies against these attacks. Through comprehensive experiments under rubric-based grading settings, we demonstrate that current LLM-based AG systems remain highly vulnerable to PI attacks. We hope that our findings raise awareness of this emerging threat and motivate future research toward secure, robust, and trustworthy LLM-based educational systems.

---

### Constitutional On-Policy Safe Distillation

Authors: Ming Wen, Yuxuan Liu, Kun Yang, Yunhao Feng, Zhuoer Xu, Yuhao Sun, Shiwen Cui, Xiang Zheng, Yi Liu, Xingjun Ma, Yu-Gang Jiang
Published: 2026-06-02
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.03089v3

On-policy self-distillation (OPSD) has emerged as an efficient post-training paradigm by using a teacher conditioned on privileged information to provide dense token-level supervision. Prior work has shown that OPSD can collapse in verifiable reasoning tasks, while safety alignment differs in that it is guided by high-level constitutions rather than explicit target answers. However, pilot studies reveals that safety OPSD nonetheless suffers from severe collapse, where constitutional conditioning contracts the teacher distribution toward short and overly conservative responses and Reverse KL further amplifies this contraction into reduced expressiveness. We formalize this effect as geometric leakage under safety boundaries in a non-orthogonal semantic space, where safety pressure transfers into the expressiveness dimension. Based on this analysis, we propose Constitutional On-Policy Safety Distillation (COPSD), which first calibrates the teacher through a Cross-SFT cold-start and then performs constitution-conditioned on-policy distillation. Experiments across 3 multimodal large language models (MLLMs) on 12 safety and general benchmarks show that COPSD improves both safety and helpfulness over baselines while reducing the safety tax on general reasoning.

---

### Constitutional On-Policy Safe Distillation

Authors: Ming Wen, Yuxuan Liu, Kun Yang, Yunhao Feng, Zhuoer Xu, Yuhao Sun, Shiwen Cui, Xiang Zheng, Guoyu Wang, Xingjun Ma, Yu-Gang Jiang
Published: 2026-06-02
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.03089v2

On-policy self-distillation (OPSD) has emerged as an efficient post-training paradigm by using a teacher conditioned on privileged information to provide dense token-level supervision. Prior work has shown that OPSD can collapse in verifiable reasoning tasks, but safety alignment differs in that it is guided by high-level constitutions rather than explicit target answers, making it a natural setting to revisit dense distillation. However, our pilot study show that safety OPSD still suffers from severe collapse: constitutional conditioning contracts the teacher distribution toward short and overly conservative responses, and Reverse KL further amplifies this contraction into reduced expressiveness. We formalize this effect as geometric leakage under safety boundaries in a non-orthogonal semantic space, where safety pressure transfers into the expressiveness dimension. Based on this analysis, we propose Constitutional On-Policy Safe Distillation (COPSD), which first calibrates the teacher through a Cross-SFT cold-start and then performs constitution-conditioned on-policy distillation. Experiments on 12 benchmarks show that COPSD achieves a consistently stronger safety--helpfulness trade-off than baselines while substantially reducing the safety tax on general reasoning ability.

---

### Learning to Solve, Forgetting to Retain: Correct-Set Turnover in RLVR

Authors: Chuanyu Qin, Chenxu Yang, Qingyi Si, Naibin Gu, Peng Fu, Zheng Lin
Published: 2026-06-02
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.03087v1

Reinforcement learning with verifiable rewards (RLVR) improves the ability of large language model, yet headline accuracy gains often conceal a hidden cost: previously solved problems quietly become unsolvable as training proceeds. We frame this phenomenon as \emph{correct-set turnover}, representing the coupled dynamics of solution acquisition and regression over the mastered set. Under this view, retention becomes an explicit optimization target alongside acquisition. We analytically and empirically establish the \emph{repair-window principle}: the cost of restoring a regressed prompt grows sharply with review delay, defining a low-cost window that standard RLVR pipelines fail to exploit. To address this, we propose \textbf{\method{}}, a retention-aware review mechanism that tracks mastered prompts and periodically reintroduces them to \textbf{remind} the model of previous solutions. By utilizing pre-rollout batch replacement, \method{} incurs zero additional rollout overhead. Evaluated across 20 benchmarks spanning image-text, video, and text-only tasks with Qwen3-VL and Qwen2.5-Math, \method{} consistently improves performance over GRPO, DAPO, and replay baselines, demonstrating robust generalizability across modalities and algorithms.

---

### Multi-component Causal Tracing in Large Language Models

Authors: Zirui Yan, Dennis Wei, Dmitriy A. Katz, Prasanna Sattigeri, Ali Tajer
Published: 2026-06-02
Categories: cs.LG, cs.CL
arXiv: https://arxiv.org/abs/2606.03085v2

Causal tracing systematically intervenes on a large language model's (LLM's) internal representations to uncover and quantify the causal pathways linking specific inputs or computations to specific metrics of interest, quantifying the LLM's behavior. Building on previous single-component or single-layer studies, this paper presents a unified framework for causally tracing multiple components simultaneously. This framework systematically identifies the subsets of components (e.g., attention heads and multi-layer perceptron neurons) most critical to a desired target performance metric (e.g., accuracy and fairness). This is achieved by incorporating flexible interventions applied to a wide range of desired metrics. To address the combinatorial complexity of the multi-component problem, an efficient algorithm is designed that leverages soft interventions and a carefully designed metric transformation, converting the combinatorial search problem into a continuous one that can be solved efficiently under proper constraints, thereby generating proper binary decisions for selecting components. Experimental results demonstrate that the proposed method efficiently identifies subsets of the model's components that have a high impact on the target metric, outperforming existing baseline approaches. Our code is available at https://github.com/ZiruiYan/multi-component-causal-tracing.

---

### Hierarchical Federated Learning with Dynamic Clustering and Adaptive Regularization for Robust Infrastructure Inspection

Authors: Yuhu Feng, Keisuke Maeda, Takahiro Ogawa, Miki Haseyama
Published: 2026-06-02
Categories: cs.CV
arXiv: https://arxiv.org/abs/2606.03084v1

The deployment of data-driven computer vision models for structural health monitoring (SHM) is heavily constrained by the data silo dilemma due to stringent privacy and security regulations. While federated learning (FL) offers a privacy-preserving collaborative alternative, its application to nationwide infrastructure networks is severely hindered by the challenge of ``double heterogeneity'': macro-level physical divergence across disparate structural types and micro-level statistical imbalances within local datasets. To overcome this challenge, this paper proposes a novel hierarchical federated learning framework. The framework orchestrates a synergistic two-tier optimization strategy. At the macro-level, a dynamic gradient-based clustering mechanism autonomously aggregates distributed clients into specialized expert groups based on their structural degradation trajectories, circumventing the need for prior geographical metadata. Concurrently, at the micro-level, an intra-cluster Dynamic Region-Adaptive Proximal Regularization (DRAPR) module computes a real-time statistical Non-IID Intensity Score for each client. By adaptively modulating a proximal penalty based on local label skewness and gradient divergence, DRAPR effectively calibrates local updates, mitigates client drift, and prevents the catastrophic forgetting of minority damage classes. Comprehensive evaluations on a large-scale, real-world structural inspection dataset demonstrate that the hierarchical integration of macro-clustering and micro-regularization successfully neutralizes dual-level heterogeneity, yielding highly robust and specialized diagnostic models for complex infrastructure inspection.

---

### DELTAMEM: Incremental Experience Memory for LLM Agents via Residual Trees

Authors: Haoran Tan, Zeyu Zhang, Zhicheng Cao, Rui Li, Xu Chen
Published: 2026-06-02
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.03083v1

Large Language Model (LLM)-based agents increasingly rely on memory to learn from experiences over continual interactions. However, storing experiences as independent, flat units leads to substantial redundancy and retrieval conflicts, as similar episodes repeat overlapping content and subtle scene variations cause retrieved memories to offer contradictory guidance. To address this, we introduce residual experience, positing that newly acquired experience is often an incremental variation of existing knowledge. We propose DeltaMem, a framework that organizes experience memory into two independent residual trees, one storing goal-conditioned task experience as reusable skills and another for scene-level environment knowledge. Each tree uses a root node for generalized base experiences and incremental delta nodes for subsequent variations, allowing related experiences to share a common foundation without duplication. For retrieval, a failure-penalized similarity scan locates the best match, reconstructing the full experience via root-to-match chain composition. An autonomous consolidation mechanism distills high-frequency paths into new root nodes, enabling the trees to self-organize from general heuristics to specialized variants. Experiments across diverse interactive environments show that DeltaMem consistently outperforms existing baselines. To facilitate future research, we release the code at https://github.com/import-myself/DeltaMem.

---

### Regret Pre-training: Bridging Prior and Posterior Views for Enhanced Knowledge Grounding

Authors: Mingkuan Zhao, Xiayu Sun, Wentao Hu, Suquan Chen, Jiaxuan Li, Xiaoyan Zhu, Xin Lai, Jiayin Wang
Published: 2026-06-02
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.03080v1

Causal language models factorize sequence probabilities using only preceding context, leaving future information unexploited during training despite its availability in the training data. This paper introduces Regret Pre-training, a self-supervised framework grounded in the Learning Using Privileged Information (LUPI) paradigm. The framework employs a dual-view architecture in which a single model generates both a causal Student distribution and a future-conditioned Teacher distribution. The training objective augments standard language modeling with a regret loss that minimizes the KL divergence from teacher to student, transferring future-aware signals to the causal representations. We investigate two teacher configurations on the OLMoE-1B-7B architecture:LocalRegret, which extends attention by one future token, andGlobalRegret, which conditions on bidirectional context with the target position masked. Experiments on nine downstream tasks following 4 billion tokens of training demonstrate that both configurations consistently outperform the baseline. On average,GlobalRegret andLocalRegret achieve 33.9% and 32.2% accuracy respectively, surpassing the baseline's 30.2%. Most notably,GlobalRegret improves BoolQ performance by 18.1 percentage points (61.0% vs 42.9%). The framework introduces no additional parameters and requires only one extra inference-mode forward pass per training step.

---

### G^2C-MT: Graph-Guided Context Selection for Document-Level Machine Translation

Authors: Baijun Ji, Zixuan Zhou, Xiangyu Duan, Yu Liu, Longbo Sun, Rupu Wei, Bohong Zhao
Published: 2026-06-02
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.03078v1

Effective document-level machine translation (DocMT) requires capturing long-range discourse dependencies. Recent work has explored retrieval-based and discourse-aware context selection. However, these approaches often lack an explicit mechanism for modeling structured discourse dependencies between distant paragraphs in a document. In this paper, we propose G^2C-MT (Graph-Guided Context for Machine Translation), which views DocMT context selection as a structured path discovery problem on a lightweight discourse graph, rather than retrieving unstructured context sets or relying on expensive LLM-based discourse modeling. In detail, we represent each paragraph as a node and model the relationship between each pair of nodes, considering their semantic similarity, adjacency, and keyword overlap. Furthermore, we propose a depth-biased random walk over the graph to sample a backward context path for each target paragraph. The context path will be used to prompt a large language model (LLM) for translation. This framework naturally supports multi-path context sampling, which can improve robustness by aggregating diverse translation candidates for discourse-ambiguous inputs. Experiments conducted across various domains show that G^2C-MT outperforms strong baselines on multiple LLMs, including DeepSeek-V3, Gemini-2.5-Flash-lite, and the Qwen-2.5/3 series.

---

### Libra: Efficient Resource Management for Agentic RL Post-Training

Authors: Kaiwen Chen, Xin Tan, Jingzong Li, Zhi Zhou, Cen Li, Jiang Liu, Meng Jie, Jiazhi Jiang, Hong Xu
Published: 2026-06-02
Categories: cs.LG, cs.AI, cs.DC
arXiv: https://arxiv.org/abs/2606.03077v3

Reinforcement learning (RL) has emerged as a standard post-training paradigm for shaping large language models (LLMs) into capable agents. In agentic RL, the rollout stage generates trajectories while invoking tools, producing long-tailed and non-stationary workloads that expose two fundamental challenges. First, due to the long-tailed response distribution, a small fraction of trajectories dominates rollout makespan.Second, rollout and training differ in their compute patterns, memory demands, and sensitivity to sequence length. As the policy evolves, shifts in the workload distribution further change their relative resource demands, making it difficult to maintain balanced execution across the two stages. We present Libra, an adaptive runtime for agentic RL post-training with two complementary components: (1) intra-stage scheduling via a Causality-Guided Bucket Scheduler that routes requests across execution buckets with different parallelism configurations, reducing delays from rollout stragglers; and (2) cross-stage coordination that dynamically reallocates workers between rollout and training as the workload changes. It moves workers between the two stages through a non-blocking protocol without interrupting ongoing training. Evaluated on a 48x NVIDIA A800 GPU cluster and a 160x Ascend 910B3 NPU cluster across three agentic benchmarks, Libra achieves up to 4.2x higher throughput and up to 2.7x faster reward convergence

---

### Libra: Efficient Resource Management for Agentic RL Post-Training

Authors: Kaiwen Chen, Xin Tan, Jingzong Li, Hong Xu
Published: 2026-06-02
Categories: cs.LG, cs.AI, cs.DC
arXiv: https://arxiv.org/abs/2606.03077v2

Reinforcement learning (RL) has emerged as a standard post-training paradigm for shaping large language models (LLMs) into capable agents. In agentic RL, the rollout stage generates trajectories while invoking tools, producing long-tailed and non-stationary workloads that expose two fundamental challenges in resource management. First, due to the long-tail distribution, a small fraction of trajectories dominates rollout makespan. Second, rollout and training are subject to cross-stage imbalance, as they exhibit strong asymmetry in compute patterns, memory demands, and sensitivity to sequence length. Compounding this asymmetry, the sequence length distribution drifts continuously as the policy evolves, rendering any static resource split progressively suboptimal. We present Libra, a resource management system to address both challenges via two core mechanisms. The first is a global resource planner that jointly optimizes GPU allocation across rollout and training clusters. It leverages an elastic hybrid pool to enable lightweight, non-blocking worker reallocation between stages. The second is a causality-driven multi-level feedback queue (C-MLFQ) scheduler, which routes requests to heterogeneous rollout buckets based on causal signals derived from tool-return outcomes, rather than relying on fragile length predictions. Evaluated on 48 A800 GPUs, Libra achieves up to 3.0x higher throughput and converges up to 2.5x faster in reward compared to the baselines.

---

### TGV-KV: Text-Grounded KV Eviction for Vision-Language Models

Authors: Jizhihui Liu, Ruizi Han, Miao Zhang, Rui Shao, Xuebo Liu, Weili Guan, Yaowei Wang
Published: 2026-06-02
Categories: cs.CV
arXiv: https://arxiv.org/abs/2606.03075v1

Vision-Language Models (VLMs) inherit the auto-regressive generation paradigm and cache the keys and values (KV) of all previous tokens to accelerate inference, resulting in memory consumption that scales linearly with context length. This issue is particularly pronounced in VLMs due to substantial redundancy in the visual modality. Although KV cache eviction approaches can effectively reduce inference memory, they often incur significant performance degradation in VLMs, as most are designed for language models and overlook the inherent gap between text and vision. By systematically analyzing the modality gap in VLMs in this work, we argue that the importance of visual information should be grounded in textual guidance and accordingly propose a Text-Grounded KV Eviction method for VLMs (TGV-KV). TGV-KV comprises three submodules: (1) Text-Vision Budgeting (TVB) assigns budget to each layer based on the mutual information interaction. (2) Text-Weighted Ranking (TWR) assesses the priority of text and ranks vision importance based on weighted text-image attention. (3) Text-Prioritised Retention (TPR) policy strategically preserves text KV to avoid acute information loss. We evaluate TGV-KV across five models with different sizes and architectures, showing that TGV-KV preserves 99.2% full-KV accuracy on the VizWiz-VQA task with LLaVA-NeXT and boosts end-to-end throughput by 52.6% with an extreme retention budget of 5%. Code is available at https://github.com/Danielement321/TGV-KV.

---

### RMPrior: Bridging Propagation Priors and Diffusion Refinement for Efficient Radio Map Construction

Authors: Zixuan Guo, Xiucheng Wang, Nan Cheng
Published: 2026-06-02
Categories: cs.LG, eess.SY
arXiv: https://arxiv.org/abs/2606.03074v1

Diffusion models achieve high-fidelity radio map construction through iterative denoising, yet their sampling cost limits practicality in dynamic wireless systems where radio maps must be refreshed repeatedly. Meanwhile, classical propagation models encode valuable scene-level knowledge that standard diffusion inference discards entirely by initializing from pure Gaussian noise. This paper bridges propagation priors and diffusion refinement through a mid-start sampling strategy. A matched propagation prior is perturbed to an intermediate diffusion timestep, and the pretrained diffusion backbone executes only the remaining reverse steps, focusing computation on multipath-aware refinement rather than full reconstruction from noise. We provide theoretical analysis establishing an upper bound on the initialization gap, a sufficient condition under which truncation improves reconstruction fidelity, and a formal characterization of prior-quality sensitivity under aggressive truncation. Experiments on IRT4HighRes show that, at $P_{\text{start}}=0.5$, the proposed method achieves a $2.01\times$ speedup while simultaneously improving NMSE, RMSE, SSIM, and PSNR over the full-step baseline. A prior-quality ablation across three propagation models of different fidelity confirms that reconstruction quality tracks prior quality, with the sensitivity amplified under shorter reverse trajectories, consistent with the theoretical predictions. These results also suggest that mid-start reconstruction quality can serve as a proxy for ranking the scene-level fidelity of different propagation models.

---

### Efficient Hyperparameter Optimization for LLM Reinforcement Learning

Authors: Minping Chen, Bowen Xiao, Du Liang, Chuxuan Zeng, Zeyi Wen
Published: 2026-06-02
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.03073v1

Reinforcement learning (RL) for large language models (LLMs) is highly sensitive to hyperparameter configurations, making hyperparameter optimization (HPO) essential yet computationally expensive. Existing multi-fidelity HPO methods remain inefficient for LLM RL due to the massive model scale and resource-intensive training cycles. In this paper, we propose Joint Fidelity Hyperparameter Optimization (JF-HPO), which simultaneously adapts both model size and training budget as fidelity. JF-HPO is empowered by: (i) it leverages a small proxy model of the target LLM for efficient training and evaluation in each HPO trial; (ii) it integrates carefully designed early-stopping strategies based on training dynamics; (iii) it introduces an efficient checkpointing mechanism to eliminate redundant computations. Compared with existing HPO methods, JF-HPO significantly improves the computational efficiency of each trial (up to 14.9 times), while achieving better or competitive predictive accuracy under the same time budget. Notably, compared with utilizing hyperparameter configurations from the VeRL Recipe, JF-HPO delivers performance improvements ranging from 5.8% to 111.6%.

---

### ASymPO: Asymmetric-Scale Policy Optimization for Asynchronous LLM Post-Training Without Behavior Information

Authors: Zehua Liu, Yuxuan Yao, Xiaojin Fu, Tao Zhong, Mingxuan Yuan
Published: 2026-06-02
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.03070v3

Asynchronous reinforcement learning can improve language-model post-training throughput by decoupling response generation from policy optimization, but stale responses introduce distribution drift. Standard behavior-corrected methods control this drift with behavior-policy probabilities, importance ratios, or clipping, which requires token-aligned, versioned, and numerically consistent behavior log-probabilities across rollout and learner systems. We ask whether asynchronous group-relative RL can instead be stabilized using only current-policy probabilities. We identify a scale-imbalance failure mode: when stale responses are evaluated under the current policy, positive and negative loss terms can appear at different negative log-probability scales, so zero-sum advantages no longer imply balanced loss contributions. We propose Asymmetric-Scale Policy Optimization (ASymPO), which normalizes each response's token loss by its current average token negative log-probability. ASymPO requires no behavior-policy probabilities, restores response-level zero-sum balance, and preserves a nonzero learning signal. We also introduce Scaled Policy Optimization (SPO), a fixed negative-scaling baseline, and evaluate both current-policy-only objectives in asynchronous mathematical reasoning post-training.

---

### ROBUST-WT: Robust Uncertainty-aware Segmentation Transform via Whitening and Training Enhancements

Authors: Aqsa Naseer, Maryam Bibi, Syeda Samiya Urooj, Muhammad Khurram Shahzad
Published: 2026-06-02
Categories: cs.CV, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.03069v1

Generalized segmentation of medical images prevents performance degradation when different imaging devices and clinical protocols are used across multiple domains. The Whitening Transform-based Probabilistic Shape Regularization Extractor (WT-PSE), published in IEEE Transactions on Medical Imaging in 2024, addresses this challenge by employing feature decorrelation and Wasserstein distance-based knowledge distillation to achieve robust cross-domain segmentation. This study systematically examines improvements to the WT-PSE learning framework. Four limitations in the original implementation are identified: limited training augmentations that fail to simulate real scanner variations, reliance on per-pixel binary cross-entropy loss that is sensitive to edge noise, the absence of a scheduled loss weighting strategy that may destabilize early training, and the lack of ablation switches for controlled scientific comparison. To address these issues, we propose four enhancements: (1) domain-adaptive augmentation including random erasing, gamma correction, and salt-and-pepper noise; (2) a hybrid BCE and Dice loss function for improved edge-aware segmentation under noisy conditions; (3) a curriculum-based Dice weight scheduling strategy; and (4) command-line control flags for systematic ablation studies. Experiments on the fundus optic disc segmentation benchmark demonstrate that the improved pipeline achieves a final epoch optic-disc Dice score of 0.956 and an ASD score of 13.31, outperforming the baseline epoch-5 Dice score of 0.939. These results indicate that training-level improvements can provide consistent performance gains without modifying the underlying WT-PSE architecture.

---

### Learn When and Where to Connect: Adaptive Virtual Nodes for Dynamic Message Passing on Graphs

Authors: Jaejun Lee, Joyce Jiyoung Whang
Published: 2026-06-02
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.03068v1

While Virtual Nodes (VNs) are often utilized in Message Passing Neural Networks (MPNNs) to facilitate effective message passing, existing VN-based methods have limitations, such as constraining all nodes to connect to the same number of VNs, fixing the connections before applying MPNNs, and connecting a node to a VN independently of the other nodes that connect to the same VN. We propose MAVN, an end-to-end differentiable MPNN framework that allows non-constrained connections between nodes and VNs and dynamically introduces VNs on demand in response to evolving node representations across layers. Specifically, MAVN learns to adaptively determine when (at which layer) and where (to which nodes) to introduce and connect VNs based on the relative importance of connections. From a pool of candidate VNs, MAVN selects the necessary VNs in each layer, where each selected VN is connected to a nonempty subset of nodes, guided by a dual-perspective scoring mechanism that jointly captures the nodes' preferences for VNs and the VNs' preferences for nodes. We theoretically prove that for any node-VN connectivity pattern, there exists a set of MAVN's parameters that can simulate the pattern. Experiments on nine real-world datasets demonstrate that MAVN consistently improves the performance of backbone MPNNs, achieving up to 46.5% improvement over the backbones and outperforms the baselines.

---

### Trajectory-Aware Node Contributions and the Limits of Static Controllability

Authors: Valentina Kuskova, Dmitry Zaytsev, Michael Coppedge
Published: 2026-06-02
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2606.03067v2

A recurring data mining task in complex networks is to determine how individual nodes contribute to system behavior. Existing approaches rely on either static-graph centralities or control-theoretic quantities such as controllability Gramians, which assume linear, time-invariant dynamics. Estimated systems, however, are typically nonlinear and time-varying. We define "emergent contribution (EC)," a finite-horizon measure of a node's dynamical leverage: the metric-weighted energy of its impulse response accumulated along the system trajectory. Computed from the Jacobians of any differentiable model, EC is estimator-agnostic and reduces exactly to average controllability in the linear, time-invariant limit. Our contribution is a characterization of when the two measures agree and diverge. Using a controlled synthetic family with known ground-truth contribution, we construct a phase diagram spanning nonlinearity, regime structure, persistence, and perturbation amplitude. EC and average controllability agree under static or smoothly drifting dynamics and both track ground truth. Divergence emerges under persistent regime switching, is strongest under persistent sign reversal, and disappears when the sign reversal is removed. At extreme perturbation amplitudes, both measures degrade, identifying the limits of local linearization. We place five estimated real systems from several domains within this phase space. Their placement serves as a diagnostic of when EC provides information beyond static controllability and therefore justifies its additional computational cost. On one panel examined in depth, a twenty-seed retraining ensemble reveals a robust variance--leverage dissociation: nodes whose perturbations propagate widely despite low within-system variance, which is not recovered by static centralities nor variance-based summaries.

---

### CORE: Conflict-Oriented Reasoning for General Multimodal Manipulation Detection

Authors: Jinjie Shen, Yaxiong Wang, Yujiao Wu, Lechao Cheng, Tianrui Hui, Nan Pu, Zhihui Li, Zhun Zhong
Published: 2026-06-02
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.03066v1

The rapid rise of generative AI has made multimodal fake news increasingly realistic and pervasive, posing severe threats to public trust and social stability. Existing detection methods rely heavily on manipulation-specific models and large-scale labeled data, resulting in poor generalization to emerging manipulation types. We observed that the essence of manipulated misinformation lies in its intrinsic conflicts, \textbf{i.e.,} semantic or physical inconsistencies either across modalities or with common world knowledge. Inspired by this observation, we propose \textbf{C}onflict-\textbf{O}riented \textbf{RE}asoning (\textbf{CORE}) framework, an effective paradigm that learns to endows multimodal large language models (MLLMs) with explicit conflict-capturing capability. To this end, CORE first constructs the Conflict Attribution Corpus (CAC) with fine-grained annotations of conflict factors and sources, providing essential data support for subsequent conflict perception training. By performing conflict-oriented representation enhancement and reasoning based on CAC, CORE achieves robust and generalizable conflict detection, effectively and rapidly adapting to unseen manipulation types with a few samples or in even zero-shot settings. Extensive experiments demonstrate that CORE surpasses state-of-the-art models. The dataset and code are publicly available at https://github.com/shen8424/CORE.

---

### ZX-Calculus:Trace-Indexed Dependent Types and Epistemic Semantics

Authors: Peng Chen
Published: 2026-06-02
Categories: cs.LO, cs.CL
arXiv: https://arxiv.org/abs/2606.03063v1

We propose ZX-Calculus (Knowledge Evolution Calculus), a conservative extension of Martin-Lof Dependent Type Theory (MLTT) integrating trace-indexed types, presheaf non-monotone semantics, and constructive AGM belief revision. A Coq mechanisation accompanies the paper (34 complete proofs; zero admits for the two central results). (I) Trace types. FinTrace(s0,sn) is an inductive family of typed execution traces. FinTrace and Star(Step) are isomorphic as path types but not judgementally equal; TraceElim exposes the event label e:Event explicitly, giving a more ergonomic interface for event-driven induction. We prove the Trace-Reachability Correspondence, Deterministic Replay, and a canonicity framework via reducibility candidates with a Transport Lemma (RC-elim deferred; all other Core results are Coq-verified). (II) Sheaf semantics. Trace-indexed propositions are contravariant sheaves over the free trace partial-order category Tf. A Separation Theorem (explicit countermodel) distinguishes proof-theoretic monotonicity from semantic non-monotonicity. The term model is an initial CwF (syntactic universal property, not classical completeness). (III) AGM belief revision. We give an explicit constructive partial meet contraction algorithm verified against (C1)-(C4). All eight AGM postulates (R1)-(R8) are theorems. Proofs of R7 and R8 use the Disjunctive Entrenchment Lemma, given a self-contained constructive derivation. (IV) Integration. B^AGM fails the sheaf composition law BP-comp for sequential revision (explicit countermodel, Coq-verified). We introduce Single-Step Revision Systems (SSRS), prove B^AGM is a valid SSRS (Coq-verified), and show this suffices for trace morphisms, retraction characterisation, and revision witnesses. The BP-comp failure reveals a fundamental tension between path-dependent belief revision and functor consistency, not previously identified.

---

### Brief Announcement: Generative Markov Model for Distributed Computing Systems

Authors: Alfreds Lapkovskis, Ali Beikmohammadi, Sindri Magnússon, Praveen Kumar Donta
Published: 2026-06-02
Categories: cs.DC, cs.AI, cs.LG, cs.NI, eess.SY
arXiv: https://arxiv.org/abs/2606.03061v1

Emerging distributed computing paradigms, such as the computing continuum, are inherently heterogeneous, stochastic, and complex. Efficiently and effectively utilizing all available resources across the continuum demands a unified formal model of the system. To address this gap, we propose a general framework for modeling distributed computing systems as a generative Markov model, factorized over a structured system state. In our model, the state decomposes into high-dimensional variables, each further factorized over its elements, reflecting the sparse dependency structure inherent to distributed systems. This yields a tractable model enabling simulation, inference, and policy learning over otherwise intractable system states, bridging distributed computing with Markov chain theory and reinforcement learning (RL). We demonstrate our framework through a case study of collaborative AI inference, in which a dedicated server combines resources with those volunteered by service users. Our results show that centralized scheduling becomes a bottleneck at scale, while distributing computation across user devices reduces both latency and server resource consumption. These findings highlight the value of adaptive decision-making in distributed computing systems and demonstrate the framework's utility for modeling, simulation, and optimization.

---

### Rethinking Molecular Text Representations for LLMs: An Empirical Study

Authors: Arun Raja, Garrett M. Morris, Kian Ming A. Chai
Published: 2026-06-02
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.03057v1

Large language models (LLMs) are increasingly used for molecular tasks, but it remains unclear which molecular representation to use. We present a systematic benchmark evaluating LLM molecular competence across nine representations and eight chemical tasks. We benchmark 16 LLMs across five model families, including reasoning and non-reasoning variants, chemistry-specialized LLMs, and closed frontier models. Performance is strongly representation-dependent and no single representation wins across tasks, though CML is the best, followed by MolJSON, InChI, and then canonical SMILES. Explicit structured text representations (CML and MolJSON) dominate structural tasks; IUPAC dominates semantic tasks, winning molecule retrieval for all 16 LLMs; and SMILES variants are rarely optimal despite their prevalence in pretraining. Chemistry-specialized models perform well with SMILES at the cost of large degradations with structured text representations, suggesting SMILES-only evaluation rewards specialization that does not generalize. Using LLM-as-a-judge, we find that IUPAC produces the highest fraction of correct molecule generations. A mechanistic study via tokenization audits, linear probes and attention shows that representations are encoded differently inside the model; for example, structured representations require higher attention across the molecular span. Our results argue against representation-invariant evaluation and motivate task-aware representation routing for LLM-based chemistry.

---

### SkillDAG: Self-Evolving Typed Skill Graphs for LLM Skill Selection at Scale

Authors: Tong Bai, Zhenglin Wan, Pengfei Zhou, Xingrui Yu, Yang You, Ivor W. Tsang
Published: 2026-06-02
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.03056v2

As LLM agents adopt large skill libraries, selecting the right subset becomes a structural problem rather than a similarity-matching one: skills depend on, conflict with, specialize, or duplicate one another, a structure invisible to both full enumeration and embedding similarity. We present SkillDAG, which models inter-skill relationships as a typed directed graph and exposes it to an LLM agent as an inference-time, agent-callable structural retrieval interface, queried and evolved during execution rather than baked into a fixed retrieval pipeline: each search returns vector matches, typed-edge neighbors, and conflict signals, and a propose-then-commit protocol lets the agent register execution-backed edges so the graph accumulates structure across episodes. On ALFWorld and SkillsBench with MiniMax-M2.7, SkillDAG reaches 67.1% success and 27.3% reward, exceeding the strongest reported Graph-of-Skills baseline by +12.8 and +8.6 points; the advantage ports to gpt-5.2-codex, and intrinsic SkillsBench Ret@K rises from 65.5 to 78.2 under matched queries. These gains trace to isolable mechanisms: candidate ranking that stays robust as the pool grows 10x where a fixed seeding-diffusion pipeline degrades, and set-monotone online edits that enlarge ground-truth recall without evicting prior hits.

---

### ToolGate: Token-Efficient Pre-Call Control for Tool-Augmented Vision-Language Agents

Authors: Anjie Liu, Yan Song, Zhixun Chen, Ziqin Gong, Zhongwei Yu, Jun Wang
Published: 2026-06-02
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.03054v1

Tool-augmented vision-language agents can acquire external perceptual evidence through OCR, detection, segmentation, and other tools, but executing every proposed tool call is costly and sometimes unnecessary. We study the pre-call control problem: after a ReAct-style VLM agent proposes a perceptual tool call, should the call be executed, or skipped before its output enters the context? Across five benchmarks, we find that the baseline agent exhibits poor local selectivity: helpful and harmful calls occur at similar rates (11.8% vs. 9.9%), while most calls do not change the immediate forced-answer prediction. We introduce ToolGate, a lightweight external controller that predicts execute/skip decisions from trajectory text and simple structural features. Across two Qwen3-VL backbones, ToolGate reduces token cost to 64-69% of the unrestricted ReAct baseline while preserving average accuracy in cross-domain settings. With matched-domain trajectory training on Qwen3-VL-30B, it further improves average accuracy by 1.65 points. These results show that tool-augmented VLM agents benefit not only from better perceptual tools, but also from explicit control over when tool outputs are worth paying for.

---

### What Do Students Learn? A Feature-Level Analysis of Dark Knowledge

Authors: Seungu Kang, Songkuk Kim
Published: 2026-06-02
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.03052v2

Knowledge Distillation (KD) is a powerful tool for model compression, yet the precise mechanisms by which student models acquire feature representations remain underexplored. In this work, we analyze student feature learning using the Interaction Tensor framework. Our analysis reveals that effective KD acts as a regularizer that prunes low-frequency, sample-specific features, encouraging the student to rely on a compact set of highly reusable features. Crucially, we observe that the dataset-level confusion matrix contains structural information analogous to the teacher's "Dark Knowledge." Leveraging this insight, we propose Confusion Distillation (CD), a teacher-free self-distillation method that utilizes the model's own evolving )confusion patterns as dynamic soft targets. CD achieves competitive performance on ResNet-34 and ResNet-50 for CIFAR-100, outperforming existing self-distillation methods like CS-KD and PS-KD by 1.2% while offering a computationally efficient alternative to standard KD.

---

### What Do Students Learn? A Feature-Level Analysis of Dark Knowledge

Authors: Seungu Kang, Songkuk Kim
Published: 2026-06-02
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.03052v1

Knowledge Distillation (KD) is a powerful tool for model compression, yet the precise mechanisms by which student models acquire feature representations remain underexplored. In this work, we analyze student feature learning using the Interaction Tensor framework. Our analysis reveals that effective KD acts as a regularizer that prunes low-frequency, sample-specific features, encouraging the student to rely on a compact set of highly reusable features. Crucially, we observe that the dataset-level confusion matrix contains structural information analogous to the teacher's "Dark Knowledge." Leveraging this insight, we propose Confusion Distillation (CD), a teacher-free self-distillation method that utilizes the model's own evolving confusion patterns as dynamic soft targets. CD achieves competitive performance on ResNet-34 and ResNet-50 for CIFAR-100, outperforming existing self-distillation methods like CS-KD and PS-KD by 1.2% while offering a computationally efficient alternative to standard KD.

---

### FCUS-rPPG: A Fast-Converging Unsupervised Framework for Remote Photoplethysmography via Gradient Oscillation Suppression

Authors: Jiajie Li, Yu Liu, Rencheng Song, Xun Chen, Juan Cheng
Published: 2026-06-02
Categories: cs.CV
arXiv: https://arxiv.org/abs/2606.03050v1

Remote photoplethysmography (rPPG) enables non-contact extraction of blood volume pulse (BVP) signals using consumer-grade cameras. Recent unsupervised rPPG methods learn BVP representations without requiring ground-truth physiological annotations, yet their optimization is often hindered by noisy and unstable gradients, resulting in slow convergence and limited cross-domain generalization. In this paper, we propose FCUS-rPPG, a fast-converging unsupervised rPPG framework with strong generalization capability. Motivated by the observation that BVP representations exhibit both multi-spectral covariation and low-dimensional manifold structure, we design a spectrally shared backbone that facilitates BVP feature disentanglement while improving optimization efficiency. To jointly enhance convergence stability and generalization performance, we further develop a unified optimization framework operating at the gradient, loss-landscape, and feature-representation levels. Specifically, a post-verification masking mechanism filters out misleading gradients according to the weak-amplitude physiological prior of BVP signals; a perturbation-based loss landscape smoothing strategy steers optimization toward more generalizable flat minima; and a noise-aware null-space regularization constrains feature updates to the orthogonal complement of the noise subspace, thereby mitigating noise-induced representation drift. Extensive experiments on five datasets demonstrate that FCUS-rPPG requires only one training epoch, whereas existing methods typically require tens to hundreds of epochs. Notably, FCUS-rPPG consistently achieves state-of-the-art (SOTA) performance in cross-dataset evaluations. This study provides an efficient and robust solution to the real-world deployment of unsupervised rPPG. The source code will be publicly available at https://github.com/JiaJieLee/FCUS-rPPG.

---

### The Value Function Semi-Algebraic Set in Partially Observable Markov Decision Processes

Authors: Ryan A. Anderson, Guido Montufar
Published: 2026-06-02
Categories: math.OC, math.AG, stat.ML
arXiv: https://arxiv.org/abs/2606.03048v1

We study the geometry of feasible value functions in infinite-horizon partially observable Markov decision processes (POMDPs) under memoryless stochastic policies. Our main contribution is a characterization of the feasible set of value functions as a semi-algebraic set, defined by explicit polynomial inequalities determined by the transition dynamics, observation kernel, and reward structure of the POMDP. This result extends prior work for fully observable Markov decision processes, where the feasible set is known to be a polytope, to the substantially more intricate partially observable setting. In contrast to the polyhedral structure arising in MDPs, partial observability induces fundamentally nonlinear constraints, leading to a richer and more complex geometric structure. Our geometric characterization provides new insight into the landscape of policy optimization in both MDPs and POMDPs, and reveals qualitative phenomena unique to partial observability, including the emergence of isolated local maximizers of the long-term reward and their dependence on the initial state distribution.

---

### The Geometry of LLM-as-Judge: Why Inter-LLM Consensus Is Not Human Alignment

Authors: Sourabrata Mukherjee, Hamna Hamna, Kalika Bali, Sunayana Sitaram
Published: 2026-06-02
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.03043v2

LLM judges now score most open-ended NLP output, and their mutual agreement is routinely read as evidence that the scores can be trusted. That reading is unsafe: judges may agree because they capture quality, or because they share the same blind spots, and agreement statistics alone cannot tell these apart. We develop a geometric test that can. Treating each judge's scores as a vector, we measure spread, effective rank, the angle to human scores, and the judge-judge, judge-human, and human-human agreement triple for 42 judges on two community-built Indic benchmarks covering four domains and eight languages. Every comparison is reference-matched: a judge and a held-out rater are scored against the same two-rater mean, since an averaged reference otherwise flatters judges by several degrees. On subjective rubrics, judges agree with one another as much as humans do yet reach only 58-66% of human agreement and often concentrate on an axis humans do not weight. On the one rubric with a verifiable answer, most of that gap closes. Ensembles converge on the judges' shared axis rather than the human one, and training widens scores without rotating them. Inter-judge agreement is evidence of human alignment only after this check passes.

---

### The Geometry of LLM-as-Judge: Why Inter-LLM Consensus Is Not Human Alignment

Authors: Sourabrata Mukherjee, Hamna Hamna, Kalika Bali, Sunayana Sitaram
Published: 2026-06-02
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.03043v1

LMs-as-judges are now standard, yet judges agree strongly with one another while agreeing only weakly with humans. We test whether this reflects shared signal or shared bias by measuring four geometric quantities on the standard LLM-as-judge stack across four community-built Indic datasets, eight Indic languages, and 41 LLM judges: score spread, effective rank, principal angle to the human subspace, and stacked correlations among judges and humans, all with bootstrap confidence intervals. On subjective rubrics, judges use less than half the human score range ($σ_J / σ_H \approx 0.3$--$0.5$). Their evaluation axis is nearly orthogonal to the human one and noticeably further from humans than humans are from each other ($87^\circ$--$89^\circ$ versus $78^\circ$--$81^\circ$). Inter-LLM agreement exceeds LLM--human agreement ($r_{LL} \approx 0.35$ versus $r_{LH} \approx 0.27$--$0.32$). On a rubric with a verifiable factual answer, the same diagnostics fall back into the human range (axis $58.5^\circ$; $r_{LH} = 0.519$). Fine-tuning and preference optimization recover spread ($0.32 \rightarrow 1.08$) but barely move the axis (still $87^\circ$--$88^\circ$). Only post-hoc calibration on a small human-anchored set improves all four community-health rubrics together, placing a calibrated 24B Indic judge ($r = 0.184$) ahead of GPT-5.5 ($r = 0.123$), yet still short of human reliability (human-human $r = 0.474$ on the verifiable rubric). We argue that inter-LLM agreement should be considered evidence of human alignment only when a direct geometric check on the judge's score subspace passes; otherwise, the consensus reflects agreement within a collapsed subspace.

---

### RelGT-AC: A Relational Graph Transformer for Autocomplete Tasks in Relational Databases

Authors: Phillip Jiang
Published: 2026-06-02
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.03040v1

Relational databases underpin modern enterprise, scientific, and healthcare systems, yet predictive machine learning on such data remains challenging due to their multi-table, heterogeneous, and temporal structure. Relational Deep Learning (RDL) addresses this by representing databases as heterogeneous graphs and applying graph neural networks (GNNs) directly. RelBench v2 recently introduced autocomplete tasks -- a practically motivated task type where the goal is to predict an existing column value from relational context, analogous to an intelligent form-filling assistant. We propose RelGT-AC (Relational Graph Transformer for Autocomplete), extending the RelGT architecture with three targeted contributions: (1) a column masking strategy that prevents trivial solutions by masking the target column during subgraph encoding; (2) a unified task head supporting binary classification, multiclass classification, and regression autocomplete tasks within a single model; and (3) a TF-IDF text encoder that automatically detects and encodes free-text columns, recovering strong lexical signal that categorical encoders discard. Across 7 tasks spanning 3 RelBench v2 datasets (rel-trial, rel-f1, rel-stack), RelGT-AC outperforms the GraphSAGE baseline on all 3 regression autocomplete tasks and achieves up to +10 AUROC points on text-heavy eligibility tasks via the TF-IDF encoder.

---

### Will Accurate Fields Mislead Photonic Design? FromGlobal Accuracy to Port Readout

Authors: Yitian Zhang, Yonghong chen, Youming Chen, Yiyang Li, Xing Zhe, Renhe Lu, Shaolin Liao, Yuzhe Ma, Zhong Guan
Published: 2026-06-02
Categories: cs.LG, physics.comp-ph, physics.optics
arXiv: https://arxiv.org/abs/2606.03038v1

Neural field surrogates can accelerate photonic design loops, but a surrogate that looks accurate in global field error can still mis-rank candidate devices when the final decision depends on localized output-port readouts. This risk is acute in propagation-dominated MMI splitters and couplers, where port power, splitting, phase, and coupling are determined by accumulated modal interference and output-window aggregation rather than by average field similarity alone. We study this field-to-design mismatch through a Field/Mediator/Readout view that separates dense complex-field error from propagation-profile and output-window errors before port aggregation. To align the surrogate with this chain, we propose PaNO, a propagation-aligned neural operator that keeps the full-field prediction interface while organizing latent states around local boundary structure, transverse modal content, axial propagation, and cross-mode interaction. We also evaluate PaNO-R2, an output-aware feedback variant for residual field components near the port region. On a 15-wavelength tunable $3{\times}3$ MMI benchmark with 4608 held-out fields, PaNO lowers NeurOLight's port-power error from 0.2018 to 0.0739 despite slightly higher cMAE, showing that global field accuracy alone is not sufficient for design-relevant readout fidelity. PaNO-R2 attains the best cMAE, propagation-profile error, output-profile error, and port-power error, reducing NeurOLight's port-power and output-profile errors by 72.7\% and 72.5\%.

---

### TriEval: A Resource-Efficient Pipeline for LLM Bias, Toxicity, and Truthfulness Assessment

Authors: Akshatha Srikantha, Manpreet Singh, Yash Jajoo, Shyamal Lakhanpal
Published: 2026-06-02
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.03036v1

LLMs have evolved from basic chatbots to the backbone of the AI ecosystem, now widely used in healthcare, schools, and government services. The domain-wide adoption of LLMs necessitates continuous evaluation to ensure their safety and fairness. Common issues encountered after deploying LLMs include inconsistent outputs and hallucinations of incorrect information. Although numerous LLM evaluation tools exist, most are limited to testing a single parameter at a time or require massive computational resources that are not accessible to most researchers. TriEval addresses these challenges by evaluating LLM outputs across multiple parameters, including bias, toxicity, and truthfulness together, while minimizing computing resources. The pipeline is compatible with both open- and closed-source models and runs on a standard laptop without a GPU cluster. TriEval has been tested on four models: Llama 3 8B, Mistral 7B, Gemma 2 9B, and Claude Haiku. The results show clear differences between open-source and closed-source models, especially in terms of toxicity and truthfulness. TriEval is being released as open source to enable broader access for researchers with limited computational resources.

---

### Capability Advertisement as a Market for Lemons: A Trust Layer for Heterogeneous Agent Networks

Authors: Gaurav Naresh Mittal
Published: 2026-06-02
Categories: cs.MA, cs.AI
arXiv: https://arxiv.org/abs/2606.03034v1

Large language model (LLM) agents have begun to delegate work to one another. Protocols such as the Model Context Protocol (MCP) and the Agent2Agent protocol (A2A) let an agent publish what it can do and let others call it, and public registries of such agents are already appearing. These protocols assume an advertised capability is a static, truthful fact. A real agent is none of these things: its competence is probabilistic, varies with input, drifts when the underlying model is updated, and, because the agent is itself a language model, it can describe itself with complete confidence and be wrong. A caller therefore sees what an agent claims to do, not what it can do, with no principled way to tell a reliable provider from a fluent impostor. We argue these difficulties share one cause: the market for lemons. When quality is hidden and claims are cheap, good and bad providers become indistinguishable, honest reliability goes unrewarded, and the market decays toward its worst participants. Economics offers three remedies, signaling, screening, and reputation, and none are present in today's agent protocols. We make four contributions: (1) a failure taxonomy that names confident-wrong as a non-adversarial, correlated subclass of Byzantine faults that classical fault-tolerance mismodels; (2) a market-for-lemons model showing that faith-based protocols admit only a low-trust equilibrium; (3) the Trust Layer, a thin, protocol-agnostic narrow waist above MCP and A2A that adds probabilistic capability descriptors, screening, and reputation, and admits a separating equilibrium when the cost of sustaining an overclaim exceeds the gain from it; and (4) a reliability-composition bound for delegation chains with an end-to-end placement argument. The design needs no model retraining and degrades gracefully when its trust anchors are absent or corrupt.

---

### Local and Global Contraction Principles for MCMC Mixing

Authors: Alireza Daeijavad, Shahab Asoodeh
Published: 2026-06-02
Categories: cs.IT, math.ST, stat.CO, stat.ML
arXiv: https://arxiv.org/abs/2606.03033v1

We develop a contraction-based framework for proving mixing-time bounds for Markov chain Monte Carlo algorithms. The framework is built around global and local contraction coefficients of Markov kernels under the $\mathsf E_γ$-divergence with $γ\ge1$. For projected Langevin Monte Carlo on a compact convex domain, we show that Gaussian smoothing yields an explicit global contraction coefficient for the $\mathsf E_γ$-divergence. This gives a direct proof of exponential convergence to the discretized stationary distribution for general smooth, possibly non-convex potentials. The rate is explicit, accommodates arbitrary random-batch sampling schemes, and yields convergence guarantees for several divergences, including KL, $χ^2$, and Rényi divergences. For independent Metropolis--Hastings with target $π$, proposal $q$, and unbounded importance weight $w=dπ/dq$, global contraction coefficients are typically trivial. We therefore introduce a local contraction coefficient on the core $C_R=\{w\le R\}$ and prove that it controls the rejection profile on the core. This yields warm-start convergence bounds governed by the local contraction coefficient and the tail profile $H_R=π(w>R)$, recovering sharp existing moment-based convergence rates when $\mathbb E_q(w^p)<\infty$ for some $p>1$, while remaining effective in heavy-tailed regimes where no finite moment of order $p>1$ exists.

---

### The Deliberative Illusion: Diagnosing Factual Attrition and Stance Homogenization in Multi-Agent LLM Deliberation

Authors: Herun Wan, Jiaying Wu, Minnan Luo, Fanxiao Li, Ningnan Wang, Nancy F. Chen, Min-Yen Kan
Published: 2026-06-02
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.03032v1

Multi-agent LLM systems often treat consensus as evidence of successful interaction. For deliberative problems, however, reliability depends on whether agents preserve the facts and viewpoints needed to interpret an issue. We identify the deliberative illusion: discussion produces (1) factual attrition, the progressive loss of issue-critical facts, alongside (2) stance homogenization, the collapse of diverse positions toward consensus. To measure this process, we introduce DelibTrace, a framework that decomposes each issue into atomic facts, labels issue-critical ones, distributes them across agents, and tracks their survival across discussion rounds. Across ethical and news-based deliberation with three representative LLM families, multi-agent discussion erases up to 72% of issue-critical facts. This loss is consequential: retained evidence can reconstruct the issue misleadingly, final stances remain anchored in base-model priors, and a single malicious agent can inject misinformation into the shrinking shared context. These results reveal a sharper risk: agents can agree more while knowing less. We call for evaluations that measure which facts, uncertainties, and legitimate disagreements survive interaction.

---

### AUDITFLOW: Executable Symbolic Environments for Structured Financial Reporting Verification

Authors: Yan Wang, Xuguang Ai, Jaisal Patel, Xueqing Peng, Fengran Mo, Yupeng Cao, Haohang Li, Mingyu Cao, Lingfei Qian, Víctor Gutiérrez-Basulto
Published: 2026-06-02
Categories: cs.AI, cs.MA, cs.SC
arXiv: https://arxiv.org/abs/2606.03031v1

Structured financial audit verification is difficult for language-model agents because correctness depends on structured evidence rather than text alone. A model must link reported facts to taxonomy concepts, traverse calculation or dimensional relations, and recompute expected values before applying an audit rule. We propose AuditFlow, a graph-grounded multi-agent framework that separates adaptive search from deterministic verification. AuditFlow builds a symbolic environment from a static US-GAAP taxonomy graph and a dynamic XBRL filing graph, and exposes it through typed tools for fact retrieval, taxonomy traversal, numerical checking, and rule evaluation. Two junior auditors inspect each case from regulatory and evidentiary views, while a senior auditor resolves disagreements and can request further investigation. The final reports are fused through evidential aggregation to produce an audit verdict, expected value, evidence trail, and trustworthiness score. On a FinAuditing-derived FinMR sample, AuditFlow reaches 82.09% joint audit accuracy under GPT-5.5, outperforming the strongest baseline by 14.93 points. Removing deterministic checks drops accuracy to 17.91%, showing that the symbolic environment performs the verification step that the model cannot reliably replace.

---

### Conditional Hypothesis Generation for LLM-Based Text Analysis with Researcher-Specified Covariates

Authors: Paiheng Xu, Jing Liu, Wei Ai
Published: 2026-06-02
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.03029v1

A core goal of computational social science is to discover interpretable differences in how language varies across outcomes of interest, such as political affiliation or instructional quality. Recent LLM-based hypothesis generation methods describe such differences in natural language, but select for globally discriminative patterns without accounting for covariates that shape the data based on researchers' domain knowledge. When covariates are ignored, selected patterns can reflect confounds rather than differences of substantive interest. We introduce conditional hypothesis generation, a framework that incorporates researcher-specified covariates to steer hypothesis discovery toward differences that hold within relevant subgroups. Two challenges arise: the target subgroup may be underrepresented (stratum imbalance), and the direction of a difference may reverse across subgroups (sign reversal). We propose two econometrics-inspired methods: one introduces feature--covariate interactions to detect sign reversals, and the other applies within-stratum demeaning and inverse-frequency reweighting to equalize underrepresented strata. Synthetic experiments show each method outperforms global baselines in its targeted setting, and expert evaluation on two real-world datasets confirms that covariate-aware generation surfaces more useful hypotheses within relevant subgroups.

---

### SEA-LION-Embedding: Open and Reproducible Text Embeddings for Southeast Asia

Authors: Peerat Limkonchotiwat, Raymond Ng, Sarana Nutanong, Jian Gang Ngui
Published: 2026-06-02
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.03027v2

Text embeddings are fundamental to many downstream applications, making robustness important for real-world NLP. However, most recent state-of-the-art embedding models are not reproducible because they rely on closed or undisclosed training data, and they remain insufficiently robust for Southeast Asian languages. We present SEA-LION-Embedding, a fully open and reproducible text-embedding pipeline for Southeast Asian languages trained only on publicly available data, and use it to study three core factors of robust embedding design: data composition, training objective, and base encoder initialization. SEA-LION-Embedding achieves state-of-the-art results on SEA-BED while enabling systematic and reproducible analysis of robust text embeddings for the region.

---

### SEA-Embedding: Open and Reproducible Text Embeddings for Southeast Asia

Authors: Peerat Limkonchotiwat, Raymond Ng, Sarana Nutanong, Jian Gang Ngui
Published: 2026-06-02
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.03027v1

Text embeddings are fundamental to many downstream applications, making robustness important for real-world NLP. However, most recent state-of-the-art embedding models are not reproducible because they rely on closed or undisclosed training data, and they remain insufficiently robust for Southeast Asian languages. We present SEA-Embedding, a fully open and reproducible text-embedding pipeline for Southeast Asian languages trained only on publicly available data, and use it to study three core factors of robust embedding design: data composition, training objective, and base encoder initialization. SEA-Embedding achieves state-of-the-art results on SEA-BED while enabling systematic and reproducible analysis of robust text embeddings for the region.

---

### Spike-Aware INT8 Execution for Spiking Language Models on Commodity CPUs

Authors: Ting Liu
Published: 2026-06-02
Categories: cs.NE, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.03026v2

Binary spike activations allow a language-model runtime to read only active weight columns and replace multiplications by weight sums. We implement this execution strategy in C++ for an 874M-parameter spike-gated language model. Sparse projections use column-major INT8 weights, integer accumulation, and one scale application per output channel; dense projections retain row-major access and FP32 activations. In a single-thread comparison using an early checkpoint, INT8 achieves 23.31 tokens/s versus 9.82 for FP32, while reducing weight storage from 3355.2 to 1087.4 MiB. A variant using INT4 on dense projections saves a further 17.4% of storage but reduces decode throughput by 46.6%. On an AMD Ryzen 7 5800X, the final INT8 checkpoint achieves 22.63 tokens/s on one thread and 47.90 on four threads; 512-token prefill reaches 94.68 tokens/s on eight threads. A separate ARM output-head case study records higher trimmed decode-window energy metrics for two candidate-verification configurations. The results characterize how activation-specific layouts and quantized kernels support CPU deployment of a spike-gated language model.

---

### Spike-Aware C++ INT8 Inference for Sparse Spiking Language Models on Commodity CPUs

Authors: Ting Liu
Published: 2026-06-02
Categories: cs.NE, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.03026v1

Spiking language models expose activation sparsity that dense Transformer runtimes do not directly exploit. This paper studies that property from a systems perspective. Building on the SymbolicLight V1 spike-gated language model family, we implement a C++ CPU inference runtime that treats sparse binary spike states as an execution primitive rather than only applying post-hoc weight compression. The runtime combines a manifest-driven weight loader, mixed row/column memory layout, AVX2/FMA kernels, per-channel symmetric INT8 quantization, and integer-domain accumulation for spike-conditioned sparse paths. On an AMD Ryzen 7 5800X, an early scalar FP32 baseline decodes at 9.5 tokens/s. Mixed-layout AVX2 FP32 raises this to 14.7 tokens/s, and AVX2 INT8 reaches 19.9 tokens/s on the same step-30k export while reducing the weight footprint from 3.49 GB to 1.06 GB. For the available 186k-step 874M-parameter INT8 export, the C++ runtime decodes at 22.63 tokens/s in a single-thread CPU benchmark, compared with 16.31 tokens/s for TinyLlama-1.1B Q8_0, 11.26 tokens/s for Falcon3-1B Q8_0, and 9.70 tokens/s for Qwen2.5-1.5B Q8_0 under llama.cpp. Thread scaling reaches 47.90 tokens/s at four CPU threads, and 512-token prefill improves from 29.86 to 94.68 tokens/s from one to eight threads. The throughput result comes with a quality cost: the SNN reports WikiText-2 perplexity 24.80, worse than the dense baselines in the same benchmark. We frame the result as an inference-systems study for sparse language runtimes, with longer-term motivation in embodied and edge agents that may benefit from local, low-core inference near sensors and actuators. Spike-aware execution can improve CPU throughput and memory behavior for sparse spiking language models, while model quality, controlled dense training baselines, embodied-task evaluation, and measured CPU energy remain open problems.

---

### Hallucinations as Orthogonal Noise: Inference-Time Manifold Alignment via Dynamic Contextual Orthogonalization

Authors: Mingkuan Zhao, Wentao Hu, Tianchen Huang, Yuheng Min, Suquan Chen, Yide Gao, Yanbo Zhai, Shuangyong Song, Xuelong Li
Published: 2026-06-02
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.03022v1

Hallucination in Large Language Models (LLMs), characterized by the generation of content inconsistent with contextual facts or logical constraints -- remains a persistent challenge for reliable deployment. In this work, we address this issue through a geometric framework rooted in the linear representation hypothesis. We propose that hallucinations manifest as orthogonal noise relative to the semantic manifold of the residual stream. Specifically, we hypothesize that while attention heads ideally propagate information congruent with the context subspace, hallucinations arise when specific heads introduce components orthogonal to this subspace, disrupting the coherence of the latent representation. Based on this formulation, we introduce Dynamic Contextual Orthogonalization (DCO), an inference-time intervention method. DCO utilizes the input residual stream as a dynamic context anchor to perform orthogonal decomposition on attention head outputs. To distinguish between context-aligned semantic updates and divergent noise, DCO employs a layer-wise Z-score suppression mechanism that selectively attenuates outlier orthogonal components based on statistical distributions. Evaluations on Llama-3-8B and 70B across benchmarks such as XSum, NQ-Swap, and IFEval demonstrate that DCO achieves superior contextual faithfulness compared to state-of-the-art intervention baselines. Furthermore, DCO maintains high performance on knowledge-intensive tasks like TriviaQA and TruthfulQA, effectively mitigating the trade-off between hallucination suppression and parametric knowledge retention often observed in existing methods. Our findings validate the geometric interpretation of hallucinations and establish DCO as a computationally efficient approach for enforcing manifold alignment.Our code is available at https://github.com/Harry-Miral/DCO

---

### Hint-Guided Diversified Policy Optimization for LLM Reasoning

Authors: Zhiyu Cao, Kaixin Wu, Mingjie Zhong, Peifeng Li, Can Ye, Qiaoming Zhu
Published: 2026-06-02
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.03021v3

Recent developments in Large Language Models (LLMs) have showcased impressive reasoning capabilities, with Reinforcement Learning with Verifiable Rewards (RLVR) being a promising enhancement strategy. However, existing reward mechanisms are constrained to the outcome-level correctness and lack explicit signals to guide the model to consider diverse solutions. In contrast, human problem solving typically involves evaluating multiple potential approaches and selecting the most reliable solution, a cognitive process that current RLVR frameworks do not explicitly incentivize. Inspired by this, we propose Hint-Guided Diversified Policy Optimization (HDPO), allowing the model to first list all potential candidate solution outlines as hints and then select the most reliable one for further reasoning. HDPO comprises two stages of Cold Start for Structured Reasoning and Hint-Guided Diversified Reinforcement Learning to incentivize the model to generate diverse and reliable solutions following the ``propose-select-think'' trajectory. Experimental results show that HDPO effectively boosts LLM reasoning and enhances the diversity of candidate solutions as well as the LLM's ability to identify reliable solutions.

---

### Reproducibility is the New Copyleft: Defining AGI-oriented Reproducible Builds

Authors: Masayuki Hatta
Published: 2026-06-02
Categories: cs.CY, cs.AI
arXiv: https://arxiv.org/abs/2606.03019v1

Copyleft, as implemented in licenses such as the GNU General Public License, was a legal hack that used copyright to guarantee user freedom by tying the availability of source code to every act of distribution. Its normative force rested on an implicit technical premise: that source code and object code stand in a well-defined, humanly auditable, and reproducible relationship. Large language models and, prospectively, Artificial General Intelligence (AGI) systems systematically violate this premise. The artifacts jointly required to reconstruct a model -- code, data, weights, hyperparameters, toolchain, and hardware configuration -- are each subject to independent legal, technical, and economic constraints that no current open-source framework fully resolves. Sufficiently capable AI systems can also rewrite licensed source into functionally equivalent derivatives stripped of their original obligations, a form of laundering against which copyleft has no effective defense. This paper argues that a functional analogue of copyleft for AGI must be grounded not in share-alike clauses over code, but in reproducible builds: a practice guaranteeing bit-exact reconstructability from declared inputs. We review the logic of copyleft, critically examine Maffulli's Second Liberation thesis according to which AI fulfills Stallman's dream, and show that the argument collapses unless AGI systems are themselves reproducible. Drawing on the Open Source AI Definition (OSAID), the Model Openness Framework (MOF), OpenMDW, and deterministic-inference research, we define seven requirements for AGI-oriented reproducible builds. We further argue that the Model Context Protocol (MCP) and analogous AI-to-AI coupling mechanisms constitute a new dynamic linking layer for which copyleft-style licensing is ill-suited, and that Masnick's "protocols, not platforms" framework offers a more promising governance template.

---

### A Fast Screening Approach for High-dimensional Outcomes and High-dimensional Predictors

Authors: Hongju Park, Zhenyao Ye, Shuo Chen
Published: 2026-06-02
Categories: stat.ME, cs.LG, math.ST, stat.ML
arXiv: https://arxiv.org/abs/2606.03018v1

Modeling interactions among multimodal, high-dimensional data is intrinsically challenging due to ultra-high dimensionality and complex dependence structure with high level noise. Screening methods are effective for reducing dimensionality, but most existing approaches shrink only the predictor space while retaining all outcomes. In cross-modal analyses, different outcomes often select different predictor subsets, so the union remains large and the response dimension is unchanged, limiting the practical benefit of screening. This gives rise to heavy computational burdens and poor interpretability. To address these limitations, we propose a new screening framework, Graph Independence Dual Screening (GIDS), which simultaneously reduces the dimensionality of response variables and predictors. We design computationally efficient algorithms that facilitate downstream selection procedures, improving accuracy and scalability, and establish supporting theoretical results. Extensive simulation studies demonstrate that GIDS outperforms existing methods that screen only predictors. To illustrate its utility, we applied GIDS to the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, analyzing interactions between genome-wide 865,353 DNA methylation and 49,386 transcriptomic variables. GIDS reduced the feature space to approximately 9,000 CpGs and 2,000 transcripts, uncovering blockwise interaction structures: clusters of CpG sites and gene transcripts with strong associations. These findings not only improve computational tractability but also yield interpretable biological insights, highlighting coordinated regulatory mechanisms underlying Alzheimer's disease.

---

### ConTraIRL: Factorized Contrastive Abstractions for Transferable IRL

Authors: Yikang Gui, Bikramjit Banerjee, Prashant Doshi
Published: 2026-06-02
Categories: cs.LG, cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2606.03017v2

Reward transfer in Inverse Reinforcement Learning (IRL) is unreliable when policies must generalize to unseen combinations of environment dynamics and task goals. We propose Factorized Contrastive Abstractions for Transferable IRL (ConTraIRL), a framework that enables compositional reward transfer by learning decoupled latent representations of these two factors. ConTraIRL uses a dual-encoder architecture that maps observations into separate dynamics and goal latent spaces, trained with a dual contrastive objective. Temporal alignment encourages the dynamics encoder to learn goal-invariant structure, while the goal encoder captures dynamics-invariant features. This factorization supports reward inference under recombined dynamics-goal settings. Experiments on continuous control benchmarks demonstrate effective few-shot transfer to unseen dynamics-goal pairings, improving sample efficiency and reward recovery over transfer IRL baselines.

---

### ConTraIRL: Factorized Contrastive Abstractions for Transferable IRL

Authors: Yikang Gui, Bikramjit Banerjee, Prashant Doshi
Published: 2026-06-02
Categories: cs.LG, cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2606.03017v1

Reward transfer in Inverse Reinforcement Learning (IRL) is unreliable when policies must generalize to unseen combinations of environment dynamics and task goals. We propose Factorized Contrastive Abstractions for Transferable IRL (ConTraIRL), a framework that enables compositional reward transfer by learning decoupled latent representations of these two factors. ConTraIRL uses a dual-encoder architecture that maps observations into separate dynamics and goal latent spaces, trained with a dual contrastive objective. Temporal alignment encourages the dynamics encoder to learn goal-invariant structure, while the goal encoder captures dynamics-invariant features. This factorization supports reward inference under recombined dynamics-goal settings. Experiments on continuous control benchmarks demonstrate effective few-shot transfer to unseen dynamics-goal pairings, improving sample efficiency and reward recovery over transfer IRL baselines.

---