# Research Papers — 2026-06 · bundle page 200 of 260

> Full-text markdown of papers #9951–10000 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/200
[← Previous bundle](https://aipod.dev/papers/archive/2026-06/page/199.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-06/page/201.md)

### Summarization is Not Dead Yet

Authors: Dongqi Liu, Chenxi Whitehouse, Zheng Zhao, Zhuchen Cao, Jian Li, Yabiao Wang
Published: 2026-06-06
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.08000v3

The progress of large language models (LLMs) has fueled claims that model-generated summaries rival or even surpass human-written references, raising questions about whether summarization remains an open research problem. We re-examine this narrative through a multi-track evaluation covering diverse datasets and state-of-the-art LLMs, combining controlled human assessment, bias-mitigated LLM-as-Judge protocols, factuality verification against external knowledge, and corpus-level linguistic analysis. Our findings reveal a more nuanced landscape in which human references continue to demonstrate advantages in informativeness and faithfulness, whereas LLM outputs are preferred mainly for surface-level coherence and fluency. Factuality verification indicates that human references remain more reliable, particularly for claims involving reasoning or synthesis, and linguistic analysis uncovers a pattern of stylistic homogeneity across different models. These observations suggest that current LLMs have raised the floor of summarization quality, but the ceiling of their performance remains below human capabilities.

---

### Summarization is Not Dead Yet

Authors: Dongqi Liu, Chenxi Whitehouse, Zheng Zhao, Zhuchen Cao, Jian Li, Yabiao Wang
Published: 2026-06-06
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.08000v1

The progress of large language models (LLMs) has fueled claims that model-generated summaries rival or even surpass human-written references, raising questions about whether summarization remains an open research problem. We re-examine this narrative through a multi-track evaluation covering five diverse datasets and five state-of-the-art LLMs, combining controlled human assessment, bias-mitigated LLM-as-Judge protocols, factuality verification against external knowledge, and corpus-level linguistic analysis. Our findings reveal a more nuanced landscape in which human reference summaries continue to demonstrate advantages in informativeness and faithfulness, whereas LLM outputs are preferred mainly for surface-level coherence and fluency. Factuality verification indicates that human references remain more reliable, particularly for claims involving reasoning or synthesis, and linguistic analysis uncovers a pattern of stylistic homogeneity across different models. These observations suggest that current LLMs have raised the floor of summarization quality, but the ceiling of their performance remains below human capabilities.

---

### Efficient Skill Grounding via Code Refactoring with Small Language Models

Authors: Sera Choi, Wonje Choi, Saehun Chun, Daehee Lee, Jooyoung Kim, Chaeun Lee, Honguk Woo
Published: 2026-06-06
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.07999v1

Effective skill grounding is essential for deploying reusable skills in embodied agents, as even minor embodiment or environmental differences can render an entire skill incompatible. This challenge is particularly pronounced in embodied settings, where agents must operate in dynamic, partially observable environments without access to large language models (LLMs). In this setting, reliance on LLMs is impractical, while small language models (sLMs) remain insufficient for the effective skill grounding required for reliable long-horizon control. We present RECENT, a refactoring-centric agent framework that enables efficient skill grounding with sLMs by decoupling skill semantics from embodiment- and environment-specific execution binding. By representing skills as executable code, RECENT preserves the semantic intent encoded in a skill's control structure while grounding it by modifying only execution bindings through localized refactoring, rather than regenerating code from scratch. We evaluate RECENT across diverse skill grounding scenarios spanning multiple robot embodiments in dynamic environments, demonstrating robust long-horizon performance when deployed with an sLM. Across all scenarios, RECENT achieves the best performance among sLM-based Code-as-Policies (CaP) methods and matches the task performance of LLM-based CaP.

---

### Enhancing AI Interpretability with Localised Architectures

Authors: Ian Seet, Jonas Bozenhard, Simon Ostermann
Published: 2026-06-06
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.07998v3

Recent advances in generative AI, especially powerful Large Language Models (LLMs), raise concerns over the interpretability, safety and sustainability of these large and opaque AI models. The power of such architectures is derived not only from the scalability of deep neural networks (DNNs), but also massively parallel hardware such as GPU clusters. The diffuse nature of DNNs gives them great function-approximation capability when provided with sufficient training data but imposes a cost in interpretability and computational efficiency. Observing that localised machine learning (ML) models tend to be more interpretable and computationally efficient than DNNs on small datasets, we reason by analogy that similar advantages may apply to specific localised hardware ML architectures. We argue that localised architectures with lower bandwidth but higher expressivity per node have the potential to be fundamentally more interpretable than DNNs running on GPU clusters while remaining competitive for smaller datasets. We then survey the suitability of various hardware ML paradigms for implementing such localised architectures and evaluate their per-node expressivity, energy efficiency and practical maturity of the technology required.

---

### Enhancing AI Interpretability and Safety through Localised Architectures

Authors: Ian Seet, Jonas Bozenhard, Simon Ostermann
Published: 2026-06-06
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.07998v2

Recent advances in generative AI, especially powerful Large Language Models (LLMs) and Large Reasoning Models (LRMs), raise concerns over the interpretability, safety and sustainability of these large and opaque AI models. The power of such architectures is derived not only from the scalability of deep neural networks, but also massively parallel hardware such as GPU clusters. The diffuse nature of deep neural networks gives them great function-approximation capability when provided with sufficient training data but imposes a cost in interpretability and computational efficiency. Observing that localised machine learning (ML) models tend to be more interpretable and computationally efficient than deep neural networks on small datasets, we reason by analogy that similar advantages may apply to specific localised hardware ML architectures. We argue that localised architectures with lower bandwidth but higher expressivity per node have the potential to be fundamentally more interpretable than deep neural networks running on GPU clusters while remaining competitive for smaller datasets. We then evaluate the suitability of various hardware ML paradigms for implementing such localised architectures and evaluate their per-node expressivity, energy efficiency and practical maturity of the technology required.

---

### MC-PDD: Masked Corpus-Level Pretraining Data Detection for Black-Box Large Language Models

Authors: Kaixin Lan, Mu You, Tao Fang, Binkai Ou, Lidia S. Chao, Derek F. Wong
Published: 2026-06-06
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.07996v1

Pretraining is fundamental to the development of Large Language Models (LLMs), yet the opacity of pretraining data complicates model analysis and raises ethical, legal, and fairness concerns. Detecting whether specific datasets were used during pretraining is, therefore, critical. Existing state-of-the-art methods typically rely on access to model probability distributions, making them unsuitable for closed-source LLMs that provide only input-output interfaces. To address this limitation, we introduce Masked Corpus-level Pretraining Data Detection (MC-PDD), a novel method inspired by the masked language modeling paradigm. MC-PDD masks highly specific tokens in each text and prompts the LLM to predict the missing content. It then assesses whether the difference in prediction hit rates between a candidate corpus and a reference non-member corpus is statistically significant. Based on this comparison, MC-PDD determines whether the candidate texts were likely included in the model's pretraining data. Experimental results demonstrate clear and consistent differences in prediction hit rates between pretrained and unseen data across three datasets, for both open-source and closed-source LLMs. Despite operating under a stricter black-box setting, MC-PDD achieves performance comparable to existing detection methods. Our approach enables practical applications such as model auditing and data copyright verification using only standard API access. Upon acceptance, we will publicly release the code and datasets.

---

### Customer-Agent: Overcoming Context Limitations in Ultra-Long Shopping Trajectories via Tool-Augmented Agents and RLVR

Authors: Hongye Liu, Rongmei Lin, Anurag Kashyap, Hejie Cui, Ricardo Henao, Besnik Fetahu, Bing Yin
Published: 2026-06-06
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.07995v1

Understanding customer shopping trajectories is essential for enabling personalized shopping experiences. However, shopping records (i.e., customer's search, clicks, purchases, etc.) often span long time horizons over multiple years, resulting in extremely long trajectories that pose significant challenges for existing large language models (LLMs). Despite the importance of this problem, existing benchmarks are limited to short customer trajectories, while real-world trajectories from large e-commerce platforms are rarely accessible due to data privacy constraints. To address this gap, we introduce ShopTrajQA, a long-context evaluation benchmark constructed from real-world product information and simulated shopping trajectories. The dataset includes variants of up to 32k and 64k tokens, enabling systematic evaluation of model robustness under varying context lengths. Through comprehensive benchmarking of frontier LLMs, we identify critical performance gaps in reasoning over long shopping trajectory data. To address these challenges, we propose a Customer Agent Framework for ultra-long context management. Leveraging a Reinforcement Learning with Verifiable Rewards (RLVR) agentic training paradigm, our approach stores trajectories as external local files and trains the agent to autonomously retrieve and parse them through code-interpreter interactions (e.g., SQL queries), effectively bypassing the fixed in-context window constraints of LLMs. Experimental results demonstrate that our framework achieves strong performance for ShopTrajQA and shows generalization to other complex reasoning tasks.

---

### VATS: Exploiting Implicit Authority in Error-Path Injection via Systematic Mutation

Authors: Harshil Patel, Kunal Pai
Published: 2026-06-06
Categories: cs.AI, cs.CR, cs.SE
arXiv: https://arxiv.org/abs/2606.07992v1

As the Model Context Protocol (MCP) standardizes tool-calling for autonomous agents, it introduces a critical, unexamined attack surface: the error-handling loop. We hypothesize that tool error messages possess implicit authority, triggering corrective reasoning modes that bypass standard safety heuristics. We introduce VATS (Vulnerability Analysis of Tool Streams), a mutation-driven framework that systematically evolves adversarial payloads across seven structural and linguistic dimensions. Our evaluation across four frontier models, Gemini 3.1 Pro, GPT-5.5, GLM-5.1, and Qwen3-Coder, demonstrates that error-path injection triples the success rate of standard indirect prompt injection (IPI), achieving up to 100% compliance in controlled evaluations. We isolate structural positioning (sandwiching instructions within error context) as the most effective exploit vector across all tested models. While we find that production framework guardrails can mitigate these vulnerabilities, the inherent susceptibility of the model layer poses a systemic risk to bespoke agentic workflows.

---

### PAFO: Pareto Fairness Optimization for Personalized Reward Modeling

Authors: Xiaoyan Zhao, Haoting Ni, Yang Zhang, Chunyuan Zheng, Haoxuan Li, Fuli Feng
Published: 2026-06-06
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.07988v1

Large language models (LLMs) increasingly rely on reward models to align their outputs with diverse user preferences. While personalized reward models aim to capture such heterogeneity, they are often trained on imbalanced user preference data and may therefore favor users whose preferences are more common in the training population. In this paper, we identify this failure mode as personalized reward bias, where reward modeling quality varies systematically with preference support rate. We formulate its mitigation as a Pareto fairness problem over group utilities, aiming to improve under-served users without degrading other user groups. To this end, we propose PAFO, a Pareto fairness optimization framework for personalized reward modeling. PAFO first trains group-specialized reward models for majority and minority preference groups, then constructs conditional margin-level supervision to distill their heterogeneous preference boundaries into a single unified model. The resulting model uses group information only during training and requires no explicit group labels at inference time. Experiments on Personal-LLM and DSP show that PAFO improves both minority-group and majority-group accuracy while reducing user-level unfairness across multiple metrics, demonstrating its effectiveness for fairer LLM personalization.

---

### Inference for High-Dimensional Sparse Spectral Precision Matrices

Authors: Navonil Deb, Younghoon Kim, Sumanta Basu
Published: 2026-06-06
Categories: stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2606.07986v2

Statistical inference for the spectral precision matrix at a given frequency allows us to assess frequency-specific conditional relationships among the components of a stationary multivariate time series. Compared with classical analogues, inference in the spectral domain is more challenging due to the absence of closed-form asymptotic variance expressions for complex-valued estimators, the limited asymptotic theory in high-dimensional settings, and the presence of truncation and smoothing biases in finite samples. We construct a debiased complex graphical lasso estimator at any fixed frequency by leveraging the full likelihood structure of neighboring discrete Fourier transforms. Using asymptotic distributions for bilinear forms of stationary multivariate time series, we establish the joint asymptotic normality of the real and imaginary parts of the debiased estimator. Our main theoretical contributions include deriving a closed-form asymptotic covariance matrix for the real and imaginary parts, establishing a central limit theorem for bilinear forms of the underlying time series, and controlling smoothing and truncation biases in covariance estimation to ensure valid inference. Simulation studies demonstrate reliable coverage and improved statistical power relative to the benchmark, while maintaining false discovery rates near the nominal level. An application to real fMRI data further reveals distinct patterns of functional connectivity across selected frequencies.

---

### Inference for High-Dimensional Sparse Spectral Precision Matrices

Authors: Navonil Deb, Younghoon Kim, Sumanta Basu
Published: 2026-06-06
Categories: stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2606.07986v1

Gaussian graphical models in the spectral domain offer a principled approach for recovering conditional dependence structures in stationary high-dimensional time series. Inference on the spectral precision matrix at a fixed frequency enables tests of frequency-specific conditional associations among time series components. The problem is challenging because finite-sample discrete Fourier transforms induce truncation and smoothing biases, while the complex-valued nature of the spectral precision matrix complicates high-dimensional variance estimation, rendering methods for i.i.d. samples not directly applicable. Existing approaches do not provide full likelihood-based inference for the discrete Fourier transforms. We propose a high-dimensional inference framework for sparse spectral precision matrices using the full likelihood of neighboring discrete Fourier transforms. We construct a debiased complex graphical lasso estimator at any fixed frequency. Using asymptotic theory for quadratic forms of multivariate time series, we establish its asymptotic normality and construct entry-wise consistent covariance estimators by aggregating information across neighboring frequencies. The key theoretical contribution is the simultaneous control of regularization, finite-sample truncation, and smoothing biases, enabling valid inference. Simulation studies show reliable coverage away from zero frequency and improved detection power over the benchmark, with false discovery rates near the desired level.

---

### FMRFusion: Frequency-Aware Multi-View Representation Learning for Heterogeneous Image Fusion

Authors: Tao Zhoua, Yunlong Liu, Qinghui Chen, Zekai Zhang, Minlong Sun, Changlin Biana, Dagang Li, Wenmin Wang, Jinglin Zhang
Published: 2026-06-06
Categories: cs.CV, cs.CL
arXiv: https://arxiv.org/abs/2606.07985v1

Infrared and visible image fusion aims to generate a composite image that retains significant target information and preserves detailed textures, integrating two heterogeneous modalities. Previous image fusion methods typically adopt a single-module stacking approach to extract features from the two modalities. However, these approaches may result in incomplete learning of their distinct characteristics, thereby limiting the fusion effectiveness and constrain ing robustness in real-world heterogeneous data scenarios. To address these challenges, we propose FMRFusion, a frequency-aware multi-view representation learning network for Heterogeneous Image Fusion. A Multi-Scale Struc tural Perception Module is introduced to effectively capture discriminative structures, extracting fine-grained local structures and essential contextual information. A bilinear frequency decomposition mechanism is employed to sepa rate features into high-frequency and low-frequency components, enabling joint modeling of local details and global representations across different frequency domains. Moreover, a Cross-View Complementary Interaction is incorpo rated to explicitly model and fuse the complementary characteristics between reflected light information and radiative intensity responses, facilitating effective cross-view interaction. We further improve the Performance of the fused results by flow matching, which progressively refines the fused features by learning the transformation from coarse data to high-quality representations. Extensive experiments conducted on multiple benchmark datasets demonstrate that FMRFusion achieves superior and consistent performance across a range of fusion tasks, especially in nighttime scenarios

---

### Overcoming the Limits of Finite Difference Method; Physics-Informed Neural Network for Noisy High-Dimensional Heat Diffusion

Authors: Shreesh Bhattarai, Harish Chandra Bhandari
Published: 2026-06-06
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.07982v1

High-dimensional transient heat diffusion under noisy boundary conditions exposes a fundamental limitation of classical numerical methods: accuracy degrades catastrophically where physical noise is unavoidable. This paper presents a Physics-Informed Neural Network (PINN) framework as a systematic solution to this problem across one, two, and three spatial dimensions, establishing clear operational regimes that redefine solver selection in noisy thermal systems. Under 20% boundary noise in 3D, PINN sustains approximately 91% accuracy while Finite Difference Method (FDM) collapses to 36%, a clear decisive advantage. This is further confirmed in a physical copper thermal system, where PINN reduces boundary reconstruction error by 3.3 times under realistic noise conditions. This noise resilience is accompanied by a dimensionality-driven efficiency crossover: PINN requires fewer spacetime nodes than FDM in 3D while achieving superior accuracy, exposing the true cost of classical discretization at scale. These findings reframe solver selection: the decisive axis is not accuracy alone, but noise exposure and dimensionality jointly. When noise and dimensionality are both high, the classical solver paradigm is insufficient; this work provides the foundation to justify PINN as the operational standard in such regimes.

---

### MechLens: Late Crystallization of Factual Knowledge Explains Intervention Effectiveness in Language Models

Authors: Xueping Gao
Published: 2026-06-06
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.07978v1

Understanding where LLMs store factual knowledge is critical for hallucination mitigation. We systematically quantify Late Crystallization: factual knowledge does not gradually emerge across layers but "crystallizes" abruptly at the final layers. Across five model families (Pythia, Gemma, Qwen2.5, Llama-3.1, Mistral; 0.5--14B), 26.8%--93.4% of correct answers never enter top-10 predictions at any intermediate layer, with late emergence (>80% depth) consistent across architectures. Cross-scale (Qwen2.5-14B) and cross-benchmark (MMLU: 98.2%) results confirm generality; tuned lens rules out probe artifacts. A sentiment-classification control (0.5% for Qwen vs. 85.9% factual; 2.0% for Mistral vs. 26.8%) confirms the phenomenon is specific to factual recall. Late Crystallization yields a crystallization-guided intervention principle: CAA outperforms DoLa on moderate-crystallization models (Llama, Mistral; p<0.001), with a directionally consistent reversal on high-crystallization Qwen (+25.4% vs. +15.5% MC1, p=0.069). LayerNorm ablation shows crystallization is intrinsic to the residual stream; LN scaling (x1.2) yields +11.8% MC1 with zero inference overhead. We further reveal a Computability-Memorization Spectrum: computable knowledge crystallizes earlier (layer 22.1/28) than memorized facts (28.0/28). We release MechLens supporting five model families.

---

### PRISM: PRior-guided Imagination Sampling in world Models

Authors: Yuhai Wang, Jiawei Xia, Rongxuan Zhou, Xiao Hu, Yongliang Shi, Jing Du, Yang Ye
Published: 2026-06-06
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2606.07974v1

A learned world model provides a powerful physical intuition for evaluating future states. But its effectiveness in continuous control also depends critically on how candidate actions are generated for model-based planning. Rather than solely asking how accurately a model can simulate the future, we ask: which candidate actions are worth evaluating in the first place? Existing planners typically search arbitrarily or use expert demonstrations only to initialize a sampling mean, discarding the expert's state-conditioned confidence. Properly guiding this search requires a robust action prior, yet current approaches often rely on independent visual encoders or large-scale VLMs to obtain one. We argue that this architectural bloat is unnecessary: the exact same data - and the learned representations of the world model itself - inherently encode the agent's action intuition. We introduce PRISM, a task-agnostic framework that extracts both from a single dataset while maintaining strict architectural simplicity. Building on a standard JEPA-style latent world model, PRISM attaches a lightweight MLP directly to its frozen encoder to predict a state-conditioned Gaussian prior. At plan time, PRISM fuses this prior into the planner's sampling distribution via a precision-weighted Product-of-Gaussians update. This parameter-free, closed-form integration steers the sampling process, making the prior confident where it is and ceding control where it is not. PRISM improves success rates by 35 percentage points over vanilla world-model-based MPC on Cube and 32 percentage points on PushT, without introducing significant inference overhead.

---

### Defending Against Malicious Finetuning by Scaling Train-time Adversarial Attacks

Authors: Haoming Wen, Shi Chen, Qingyu Shi, Siyuan Liu, Minrui Luo, Jingzhao Zhang, Tianxing He
Published: 2026-06-06
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.07970v2

Malicious finetuning attacks pose a major safety threat against open-weight large language models (LLMs). However, existing alignment-stage defenses provide limited protection against strong attacks that use full-parameter finetuning. To address this issue, we propose Patcher, a novel and efficient adversarial training algorithm that alternates between an attack stage and a defense stage: the attack stage simulates many-step malicious finetuning and computes the resulting parameter displacement as an "attack vector", while the defense stage aims to preserve safe behavior under such perturbations. Compared with other adversarial training methods, a key feature of Patcher is that the attack vector can be reused across multiple defense updates, substantially reducing training costs. Theoretically, we characterize how the relative gradient estimation error is influenced by the number of attack steps during training and reuse-duration of attack vectors. Empirically, experiment results show that Patcher reduces Attack Success Rate by 67.5%, 53.4% and 54.8% on three benchmarks compared to the second-best method while preserving the model's utility. Moreover, Patcher remains effective under diverse model sizes and attack scenarios, and generalizes to prompt-based jailbreak attacks. Code is available at https://github.com/haomingwen/patcher

---

### Defending Against Malicious Finetuning by Scaling Train-time Adversarial Attacks

Authors: Haoming Wen, Shi Chen, Qingyu Shi, Siyuan Liu, Minrui Luo, Jingzhao Zhang, Tianxing He
Published: 2026-06-06
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.07970v1

Current open-weight large language models (LLMs) are prone to malicious finetuning attacks, which could compromise the safety alignment of LLMs with only a few steps of supervised finetuning (SFT) on poisoned datasets. Existing alignment-stage defenses are primarily designed to defend against attacks that use parameter-efficient finetuning methods. However, they fail to defend against stronger attacks that use full-parameter finetuning. In this paper, we propose Patcher, a method inspired by adversarial training and bi-level optimization, to combat such attacks. Patcher strengthens the simulated attack by scaling up the optimization steps in the adversarial loop, thus forcing the defender to find model parameters that are insensitive to stronger attacks. Furthermore, we propose an efficient parallel algorithm to implement Patcher, decreasing the wall-clock time of training while preserving Patcher's performance. Extensive experiments show that Patcher substantially improves the model's robustness compared to vanilla SFT alignment, and transfers to diverse attack scenarios and model sizes. Code is available at https://github.com/haomingwen/patcher.

---

### Neutrality Bites: Gender Representation in AI-Generated Animal Stories

Authors: Imani Finkley, Yuanxi Li, Melanie Walsh
Published: 2026-06-06
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.07969v1

Gender bias in AI-generated stories is a well-documented problem. While much attention has been paid to reducing or mitigating this bias, it is not always clear whether interventions produce genuinely fairer results. To investigate this issue, we examine how large language models (LLMs) handle gender assignment in a narrative context that is popular, highly ambiguous, and also known to closely reproduce human stereotypes: stories about talking animals. We prompt six leading LLMs to complete an English-language story about seven different anthropomorphic animal characters whose gender is unstated. We additionally iterate with four different narrative settings and a range of model temperatures. Across the 23.8K stories, we find that models frequently avoid gendering the animal character in the story (19% on average) or use gender-neutral language like "it" or "its" (38.2% on average). However, when gender is assigned, there is a significant masculine bias. Feminine animal characters are virtually absent, present in just 2.2% of stories vs. 40.6% that feature masculine characters. Our findings point to a broader argument: neutrality bites. In other words, models that prioritize neutrality to address social bias may actually contribute to the erasure of marginalized perspectives and identities. We suggest that alternative strategies beyond neutrality need to be pursued, such as ones that more equally distribute social possibilities across imagined subjects.

---

### RecurGuard: Runtime Monitoring for Reasoning-Token Consumption Attacks

Authors: Abid Aziz, Hafsa Binte Kibria
Published: 2026-06-06
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2606.07968v1

Reasoning-capable large language models can be induced to spend their generation budget on injected decoy tasks rather than answering the user's question, causing denial of service when no final answer is produced and denial of wallet when excess output tokens are billed. Input-side safety classifiers often miss these attacks because the injected prompts can appear syntactically benign. We build RecurGuard, a runtime monitor for detecting reasoning-chain consumption attacks when reasoning traces are exposed by the model. RecurGuard analyzes reasoning traces as they are generated and tracks three signals: recurrence rate, volume growth, and progress toward the user's query. If all three signals remain anomalous over three consecutive chunks, RecurGuard terminates generation early. We evaluate RecurGuard against OverThink and ExtendAttack across open-weight reasoning models and conduct adaptive stress tests on DS-R1-Qwen-7B. On this model, RecurGuard detects 99% of OverThink attacks and 92% of ExtendAttack instances while maintaining near-zero false positive rates on question answering, code generation, mathematics, and summarization. Adaptive evaluation reveals the limit of the defense: topical attacks retain 11.9x amplification with an approximately 50% joint miss rate, whereas full semantic evasion reduces amplification from 22.8x to 2.2x. When reasoning traces are unavailable, QDM provides a post-hoc fallback monitor based on the final output.

---

### DisCo: World Models with Discrete Camera Motion Control

Authors: Hongrui Huang, Junke Wang, Quanhao Li, Yu-Gang Jiang, Zuxuan Wu
Published: 2026-06-06
Categories: cs.CV
arXiv: https://arxiv.org/abs/2606.07967v1

Controllable video world models target interactive world exploration, where models must faithfully execute explicit action commands while preserving visual quality and temporal coherence. However, most existing approaches rely on continuous camera trajectories as action conditions, which often lead to unreliable action following, especially under complex motion sequences. In this work, we identify action representation entanglement as a key bottleneck in controllable video generation, and show that continuous camera representations lead to high feature similarity across distinct motion patterns, degrading action controllability. Based on this insight, we propose DisCo, a controllable video world model that conditions generation on a compact set of discrete action primitives to improve action separability. We further introduce DisCoBench, a comprehensive benchmark for evaluating the ability of models in short-term, long-horizon, and highly dynamic exploration scenarios. Extensive experiments demonstrate that DisCo achieves significantly more reliable action following while preserving visual quality.

---

### Zero-Shot Learning in Industrial Scenarios: New Large-Scale Benchmark, Challenges and Baseline

Authors: Zekai Zhang, Qinghui Chen, Maomao Xiong, Shijiao Ding, Zhanzhi Su, Xinjie Yao, Yiming Sun, Cong Bai, Jinglin Zhang
Published: 2026-06-06
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.07965v1

Large Visual Language Models (LVLMs) have achieved remarkable success in vision tasks. However, the significant differences between industrial and natural scenes make applying LVLMs challenging. Existing LVLMs rely on user-provided prompts to segment objects. This often leads to suboptimal performance due to the inclusion of irrelevant pixels. In addition, the scarcity of data also makes the application of LVLMs in industrial scenarios remain unexplored. To fill this gap, this paper proposes an open industrial dataset and a Refined Text-Visual Prompt (RTVP) for zero-shot industrial defect detection. First, this paper constructs the Multi-Modal Industrial Open Dataset (MMIO) containing 80K+ samples. MMIO contains diverse industrial categories, including 6 super categories and 18 subcategories. MMIO is the first large-scale multi-scenes pre-training dataset for industrial zero-shot learning, and provides valuable training data for open models in future industrial scenarios. Based on MMIO, this paper provides a RTVP specifically for industrial zero-shot tasks. RTVP has two significant advantages: First, this paper designs an expert-guided large model domain adaptation mechanism and designs an industrial zero-shot method based on Mobile-SAM, which enhances the generalization ability of large models in industrial scenarios. Second, RTVP automatically generates visual prompts directly from images and considers text-visual prompt interactions ignored by previous LVLM, improving visual and textual content understanding. RTVP achieves SOTA with 42.2% and 24.7% AP in zero-shot and closed scenes of MMIO.

---

### What Does Debiasing Really Remove? A Geometric Study of PCA-Based Gender Debiasing in Word Embeddings

Authors: Alexey Kresin, Tchifou M. Dieffi, Tomer Caspi
Published: 2026-06-06
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.07964v1

Debiasing methods based on principal component analysis (PCA) are broadly used to reduce gender bias in word embeddings used in LLMs, yet it remains unclear what aspects of bias they actually remove and how destructive this process is. These methods are based on the understanding that bias resides in a low-dimensional subspace, with the assumption that most of it can be captured by a few principal components. In this work, we conduct a systematic geometric analysis of PCA-based gender debiasing and investigate what is actually removed from the embedding space. Our experiments across multiple embeddings show that direct gender bias is primarily concentrated in the first principal component, supporting the low-rank bias hypothesis. However, associative bias measured by WEAT does not align with these principal directions and is instead spread across multiple embedding dimensions. Furthermore, as expected, we demonstrate that removing an increasing number of principal components leads to a consistent degradation of the embedding geometry, affecting semantic structure and vector relationships. These results reveal that PCA-based debiasing operates as a trade-off: while it effectively reduces certain forms of direct bias, it fails to eliminate distributed associations and introduces geometric distortion. Moreover, there is no universal optimal level of debiasing, as the balance between bias reduction and semantic preservation depends on the chosen metric and embedding. Overall, our findings suggest that bias in word embeddings is not purely low-rank and that simple subspace removal methods may be insufficient for comprehensive debiasing.

---

### Shared Latent Structures Enable Unified Backdoor Detection and Mitigation in LLMs

Authors: Omar Mahmoud, Aly M. Kassem, Thommen George Karimpanal, Buddhika Laknath Semage, Negar Rostamzadeh, Golnoosh Farnadi, Santu Rana
Published: 2026-06-06
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2606.07963v1

Backdoor attacks in large language models (LLMs) are often treated as isolated trigger-response failures, motivating defenses tailored to specific triggers or behaviors. We show this view is incomplete. Across diverse backdoor behaviors, we identify a shared latent mechanism that can be detected, causally controlled, and suppressed. Using sparse autoencoders (SAEs) on residual-stream activations, we find a small set of latent features consistently activated across jailbreaking, refusal manipulation, password-locking, bias induction, sentiment misclassification, and country-conditioned harmful advice. These features generalize across Qwen3, Gemma~3, and Llama~3.1 models from 4B to 32B parameters, and across both fine-tuning and weight-editing attacks. Through bidirectional activation steering, we show these features are causal: suppressing them reduces attack success, while amplifying them induces target behaviors on clean prompts. We further train lightweight SAE-feature classifiers that generalize zero-shot to unseen backdoors and outperform residual-stream and weight-diffing baselines. Finally, we introduce Concept Ablation Fine-Tuning (CAFT), which suppresses backdoor formation by ablating the shared latent subspace during training. Together, our results suggest that many backdoors rely on a transferable latent mechanism, enabling unified detection and mitigation.

---

### ChronoPhyBench: Do MLLMs Truly Understand the World or Merely Exploit Language Priors?

Authors: Bin Zhu, Yanhao Jia, Kexin Zhao, Jie Wang, Munan Ning, Hao Li, Yuwei Niu, Tanqing Sun, Huangchong Yan, Mingjun Pan, Xinyi Wu, Qishen Yin et al.
Published: 2026-06-06
Categories: cs.CV
arXiv: https://arxiv.org/abs/2606.07962v1

Recent advancements in Multimodal Large Language Models (MLLMs) have demonstrated remarkable proficiency in open-world reasoning and understanding. However, a critical ambiguity persists: it remains unclear whether these models genuinely synthesize cross-modal information to construct physically grounded reasoning chains, or if they merely exploit strong language priors to mask single-modality reliance, thereby hallucinating advanced multimodal capabilities. Motivated by this, and to rigorously mitigate language modality bias and shortcuts, we propose a novel multimodal Chrono}logical Physical Dynamics Reasoning Benchmark ChronoPhyBench, which unifies next state prediction with Visual Question Answering (VQA) paradigms by conditioning on historical video context and textual captions to enforce models to deduce subsequent physical states through both single image selection and the inherently more complex task of multiple frame chronological sorting. Concurrently, we construct a large-scale multimodal reasoning dataset curated using the ChronoPhyBench criteria, comprising over 10,000 long-form videos paired with meticulously annotated captions, totaling 5M tokens. Our experimental evaluations reveal a stark contrast to conclusions drawn by previous benchmarks. The capacity of current open-source models to perform physically grounded multimodal reasoning remains in its infancy. Ultimately, this work seeks to systematically stress-test the reasoning capabilities of multimodal models, quantify hallucination rates, and advance the development of Physical AI, thereby providing the community with a robust and transparent evaluation framework toward Artificial General Intelligence (AGI).

---

### Minibatch Selection for Language Models via Partition Matroid Constrained Gradient Matching

Authors: Prayas Agrawal, Prateek Chanda, Ishita Khatri, Ganesh Ramakrishnan, Bamdev Mishra, Pratik Jawanpuria
Published: 2026-06-06
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2606.07954v2

Training large language models (LLMs) on heterogeneous data requires selecting minibatches that balance convergence speed with coverage across domains. Existing methods either select samples independently within each domain or rely on computationally expensive proxy models to learn continuous domain weights. We propose PartitionSel, a cross-domain minibatch selection approach that maximizes a validation-guided gradient-matching utility under per-domain budgets encoded as a partition-matroid constraint. By coupling the per-domain budgets through a single utility, PartitionSel is designed to reduce redundancy in selections across domains. The proposed objective is weakly submodular and admits an orthogonal matching pursuit algorithm with provable approximation guarantees. Empirically, we evaluate PartitionSel for minibatch selection during the fine-tuning of Qwen2.5 and Llama-3 on MetaMathQA and Mol-Instructions. PartitionSel achieves robust gains over per-domain and domain-agnostic baselines on both benchmarks. It also reduces the number of conflicting gradient pairs within each batch, indicating that the cross-domain coupling translates into more compatible training updates.

---

### Unification of Closed-Open Industrial Detection Scenarios: New Large-Scale Benchmarks,Challenges and Baselines

Authors: Zekai Zhang, Jinglin Zhang, Qinghui Chen, Gang Li, Da Chen, Shuainan Jing, He Wang, Dagang Li, Cong Liu, Cong Bai, Shengyong Chen
Published: 2026-06-06
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.07953v1

Large-scale Visual-Language Models (LVLMs) have achieved remarkable success in natural visual tasks, yet their application to industrial defect detection remains challenging due to two fundamental limitations: (i) the scarcity of large-scale industrial datasets that cover diverse defect categories across multiple domains, and (ii) the reliance on manual prompts (points, boxes, masks) that introduce subjective noise and lack text-visual interaction for fine-grained understanding. To address these challenges, we introduce a Large-Scale Multi-Modal Industrial Open-Closed benchmark (MMIOC-1M) containing over one million samples across $14$ super-categories, $29$ industrial scenes, and $351$ defect subcategories. To our knowledge, MMIOC-1M is the first unified largest benchmark supporting both open-vocabulary and closed-set industrial detection, providing valuable pre-training data for LVLMs in industrial scenarios. Furthermore, we propose a Refined Text-Visual Prompt Network (RTVPNet) that incorporates three key innovations: (1) an expert-assisted domain projection mechanism that enables rapid adaptation of general vision models to industrial domains, (2) an energy-based sparse sampling strategy that automatically generates refined visual prompts without manual intervention, and (3) a bidirectional text-visual interaction module that enhances cross-modal semantic alignment and understanding. Extensive experiments demonstrate that RTVPNet achieves state-of-the-art performance on MMIOC-1M, LVIS, and COCO benchmarks while maintaining computational efficiency. The dataset and code are available at https://github.com/hellozzk/MMIO.

---

### From 'May' to 'Is': Certainty Distortion in Language Model Rewriting

Authors: Catarina G Belem, Shang Wu, Hongyu Yao, Mark Steyvers, Sameer Singh, Padhraic Smyth
Published: 2026-06-06
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.07951v2

Humans increasingly turn to Language Models (LMs) in ways that shape beliefs and drive decisions, including discussing, rewriting, and summarizing information from scientific articles, news, and medical reports. However, in these domains, where it often matters how confidently a claim is expressed, little is known about whether LMs faithfully preserve the degree of confidence. In this work, we investigate certainty distortion in LMs, defined as meaningful changes in expressed certainty during transformations intended to preserve meaning. We propose an LM-based evaluation metric that is consistent with population-level judgments of certainty. Using this metric, we characterize certainty distortion across different sizes and families of models in the context of scientific and medical communication tasks. Our results show that certainty distortion affects up to 75% of LM outputs and is systematically asymmetric in rewriting tasks with most LMs being 1.5-2x more likely to increase the expressed certainty than to decrease it. These effects can compound over repeated paraphrasing: in the medical domain, claude-haiku-4.5 increases certainty in 20% of examples after a single iteration, increasing to 40% after five iterations. Prompt-based interventions reduce overall certainty distortion but do not eliminate it. Together, these findings reveal a general bias toward inflating expressed certainty, with direct implications for users who rely on LMs in high-stakes domains.

---

### From `May' to `Is': Certainty Distortion in Language Model Rewriting

Authors: Catarina G Belem, Shang Wu, Hongyu Yao, Mark Steyvers, Sameer Singh, Padhraic Smyth
Published: 2026-06-06
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.07951v1

Humans increasingly turn to Language Models (LMs) in ways that shape beliefs and drive decisions, including discussing, rewriting, and summarizing information from scientific articles, news, and medical reports. However, in these domains, where how confidently a claim is expressed matters, little is known about whether LMs faithfully preserve it. In this work, we investigate certainty distortion in LMs, defined as meaningful changes in expressed certainty when semantic content is preserved. We propose an LM-based evaluation metric that is consistent with population-level judgments of certainty. Using this metric, we characterize certainty distortion across different sizes and families of models in the context of scientific and medical communication tasks. Our results show that certainty distortion affects up to 75\% of LM outputs and is systematically asymmetric in rewriting tasks with most LMs being 1.5-2$\times$ more likely to increase the expressed certainty than to decrease it. These effects can compound over repeated paraphrasing: in the medical domain, claude-haiku-4-5 increases certainty of 20\% examples after a single iteration, increasing to 40\% after five iterations. Prompt-based interventions reduce overall certainty distortion but do not eliminate it. Together, these findings reveal a general bias toward inflating expressed certainty, with direct implications for users who rely on LMs in high-stakes domains.

---

### The Easy, the Hard, and the Learnable: Confidence and Difficulty-Adaptive Policy Optimization for LLM Reasoning

Authors: Zhanke Zhou, Xiangyu Lu, Chentao Cao, Brando Miranda, Tongliang Liu, Bo Han, Sanmi Koyejo
Published: 2026-06-06
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.07950v1

RL with verifiable rewards can substantially improve LLM reasoning, yet standard GRPO-style training often treats easy, hard, and learnable questions alike through uniform sampling and weighting, leading to inefficient compute allocation. We study GRPO by tracking token log-probabilities, group-normalized advantages, and the induced token-level update weights. This reveals three recurring dynamics as training proceeds: (1) confidence inflation, (2) advantage contraction, and (3) hierarchical convergence. These findings suggest that the utility of each update depends strongly on both question difficulty and the model's current competence. Motivated by this, we propose Confidence and Difficulty-adaptive Policy Optimization (CoDaPO), which assigns each question a bounded value from rollout confidence and empirical difficulty. CoDaPO then uses this value to reweight policy updates and resample high-value learnable questions within mini-batches, thereby increasing discovery within the learnable band under a fixed compute budget. Across twelve benchmarks, CoDaPO consistently improves accuracy over existing RL methods. Our code is publicly available at https://github.com/tmlr-group/CoDaPO.

---

### Feasibility to detect rapid change and disappearance of seagrass: Lessons from nearly 80 years of vegetation change in the Ako, Seto Inland Sea, Japan

Authors: Takehisa Yamakita, Yoji Igarashi, Akira Eto, Ken Ishida, Masaaki Iiyama
Published: 2026-06-06
Categories: q-bio.PE, cs.CV, eess.IV
arXiv: https://arxiv.org/abs/2606.07949v1

This study analyses the Ako tidal flat in the Seto Inland Sea, Japan, where nearly all Zostera marina disappeared within a single year in 2025. Using aerial photographs from the 1940s onward, high-resolution satellite imagery, GRUS images (2.5-5 m), and monthly Sentinel-2 composites (10 m), we reconstructed approximately 80 years of seagrass distribution. YOLO-based segmentation using deep learning achieved high accuracy (overall accuracy >= 0.9) across these datasets; although species could not be discriminated, the models captured the major temporal dynamics in vegetation area. The long-term mean seagrass area was 6.8 ha, but values fluctuated widely, from 3.5 ha in 1974 to 41.3 ha in 1989 except 0.2 ha in 2025. Sentinel-2 composites from 2019 to 2026 revealed clear seasonality, with vegetation increasing in early summer and declining from autumn. In 2025, however, the area decreased sharply after summer and remained anomalously low throughout the winter of 2025-2026. Our results, indicating that the 2025 event was not a normal fluctuation but a rapid ecosystem shift involving the loss of the dominant canopy-forming species, most plausibly driven by regionally elevated summer water temperatures. The findings also have implications for seagrass Essential Ocean Variables (EOVs) and the State of Nature (SoN) metrics used in TNFD-aligned nature-related disclosures. Unlike forests, seagrass meadows require finer temporal resolution because both pronounced seasonality and abrupt collapse strongly influence area-based indicators. Therefore, in addition to previously noted issues such as species-level classification accuracy, we recommend that (1) baselines be defined over the longest available record and justified ecologically, (2) seasonal standardization be applied before inter-annual comparisons, and (3) years with extreme area anomalies be flagged rather than used as reference points.

---

### Poise: Position-Aware One-Instruction Skill Injection for Silent Execution on LLM Agents

Authors: Haochang Hao, Dehai Min, Zhifang Zhang, Yunbei Zhang, Miao Xu, Yingqiang Ge, Lu Cheng
Published: 2026-06-06
Categories: cs.CR, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2606.07943v2

Agent skills extend general-purpose agents, but their open format enables skill poisoning: a tampered skill can make an agent run an attacker's command while completing the user's legitimate task. Invocation alone is insufficient; the attack-specific action must complete while that task still passes its verifier. We therefore define Attack Success Rate (ASR) to require a postcondition-validated sandbox action and a passing task verifier in the same trial. Skill files expose a reliability-visibility trade-off between a preloaded but conspicuous YAML frontmatter block and a longer body, where arbitrary placement may be skipped or locally incongruent. We introduce Poise, a position-aware attack that uses context-aware generation to place exactly one benign-looking, command-bearing instruction at a structurally feasible body position. On the eligible Skill-Inject pool with codex+gpt-5.2, Poise achieves 89.3\% ASR, 28.0 points above a context-free random-placement body baseline and comparable to the 86.7\% ASR of a high-exposure YAML-only baseline. Under the SkillTester audit, four LLM judges falsely flag 74.6\% of clean skills on average across both benchmarks, while only 5.6\% of Poise variants gain a new high-risk alert over their clean counterparts. One locally plausible, command-bearing body instruction therefore matches YAML-level reliability, while the resulting poisoned skill seldom adds a new high-risk finding over its clean counterpart.

---

### POISE: Position-Aware Undetectable Skill Injection on LLM Agents

Authors: Haochang Hao, Dehai Min, Zhifang Zhang, Yunbei Zhang, Miao Xu, Yingqiang Ge, Lu Cheng
Published: 2026-06-06
Categories: cs.CR, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2606.07943v1

Agent skills provide a lightweight mechanism for extending general-purpose agents, but their open format exposes them to skill-poisoning attacks. A practically dangerous injection must stay invisible: if executing the payload derails the user's legitimate task, the resulting failure signal invites inspection of the skill. We therefore evaluate attacks by Attack Success Rate, which requires the injected payload to execute and the user's task to still pass its verifier in the same trial. Prior skill-poisoning attacks face a reliability-stealth trade-off under this lens: YAML-header injections are reliably loaded but easily inspected, whereas stealthier body injections that place explicit malicious commands in the skill prose are less reliable because out-of-context commands invite the agent's own suspicion. We introduce POISE, a position-aware attack that compresses the trigger into a single, benign-looking body instruction, placing it at a feasible position and using a context-aware generator to blend it with nearby setup or prerequisite steps. On Skill-Inject with codex+gpt-5.2, POISE achieves an 89.3% ASR, 28.0 points above a random-placement body baseline and 2.6 points above a YAML-only baseline, while retaining the stealth advantage of body placement. That stealth is the decisive margin: because legitimate skill bodies naturally require privileged tool operations, LLM scanners are hyper-sensitive, falsely flagging 74.6% of clean skills on average across four judges and both benchmarks. Blending into these false alarms, POISE causes only 5.6% of poisoned variants to gain a new high-risk alert over their clean baselines, rendering current static defenses ineffective.

---

### DAL-PCQA: Enabling Distortion-Level and Language-Driven Reasoning for Point Cloud Quality Assessment

Authors: Swarna Chakraborty, Gabriel De Castro Araújo, Syeda Tasmi Faria, Marcelo M. Carvalho, Mylene C. Q. Farias
Published: 2026-06-06
Categories: cs.CV, cs.MM, eess.IV
arXiv: https://arxiv.org/abs/2606.07938v1

Point Cloud Quality Assessment (PCQA) methods typically predict scalar Mean Opinion Scores (MOS), which quantify overall perceptual degradation but do not reveal its causes. In contrast, human observers naturally reason in terms of specific distortions such as blur, color shifts, point density changes, missing regions, and geometric deformations. To close this gap, we introduce DAL-PCQA, a distortion-aware, language-annotated dataset for PCQA. DAL-PCQA augments benchmark point clouds with multi-level distortion severity labels, discrete quality categories, and structured natural language descriptions aligned with human perception. We define a point-cloud-specific distortion taxonomy that covers both photometric and geometric artifacts. Statistical analysis reveals characteristic degradation patterns across distortion types and quality levels. To assess the utility of these annotations, we compare zero-shot and fine-tuned multimodal models for generating perceptual quality descriptions. Experiments show that distortion-aware supervision substantially improves lexical and semantic alignment with ground-truth descriptions. By enabling interpretable, distortion-level reasoning, DAL-PCQA facilitates language-driven, explainable point cloud quality assessment. The dataset is publicly available at https://github.com/swarna96/DAL-PCQA.

---

### Illusions of the Gold Standard: A Large-scale Analysis of Human Evaluation Protocols for Long-form Text Generation

Authors: Katelyn Xiaoying Mei, Yi-Li Hsu, Minjoon Choi, Zongwan Cao, Chenjun Xu, Bingbing Wen, Su Lin Blodgett, Lucy Lu Wang
Published: 2026-06-06
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2606.07936v2

Human evaluation plays a critical role in assessing the quality of generated text. However, the reliability and reproducibility of these evaluations depend on transparent and well-documented protocols -- details that are frequently missing in current practice. In this work, we conduct a large-scale analysis of human evaluation protocols for evaluating long-form generation tasks in *CL conference publications from 2023--2025, including a full manual review of 284 papers and LLM-assisted analysis for another 1.8k+ papers. We define a set of 20 reportable criteria related to reproducibility of human evaluation studies, and apply these criteria to systematically examine reporting norms and practices within the community. We find widespread under-reporting of important aspects of human evaluation study design, leading to ambiguity about what was measured and how, who contributed judgments, and how judgments should be interpreted. Based on these findings, we outline actionable recommendations to support more transparent and reproducible reporting in future research. Our analysis code and annotated dataset can be found at: https://github.com/larchlab/Illusions-of-the-Gold-Standard

---

### REACT 2026: The Fourth Multiple Appropriate Facial Reaction Generation Challenge: Personalised MAFRG and Appropriate EEG Reaction Prediction

Authors: Siyang Song, Micol Spitale, Zijian Wu, Xiangyu Kong, Cheng Luo, Cristina Palmero, German Barquero, Sergio Escalera, Michel Valstar, Mohamed Daoudi, Fabien Ringeval, Andrew Howes et al.
Published: 2026-06-06
Categories: cs.CV
arXiv: https://arxiv.org/abs/2606.07935v1

In dyadic interactions, various human facial reactions could be appropriate for responding to each human speaker behaviour. Following the successful organisation of the REACT 2023, 2024 and 2025 challenge series, a body of generative deep learning (DL) models have been developed for the problem of multiple appropriate facial reaction generation (MAFRG). This year, we propose the REACT 2026 challenge encouraging the development and benchmarking of Machine Learning (ML) models that can generate multiple personalised, appropriate, diverse, realistic and synchronised human-style facial reactions expressed by a specific human listener for responding to each given speaker behaviour. As a key of the challenge, we continuously provide challenge participants with MARS dataset introduced by REACT 2025 but additionally provide individual-level Big-Five personality labels and EEG recordings. This introduces a new one-to-many personalised facial reaction generation setting combining human expressive behavioural, affective and neurophysiological signals, which remains largely unexplored in current dyadic interaction modelling. This paper also presents the challenge guidelines and new baselines on the four proposed sub-challenges: Offline generic and personalised MAFRG as well as Online generic and personalised MAFRG, respectively, which are publicly available at https://github.com/reactmultimodalchallenge/baseline_react2026.

---

### LEGS: Laplacian-Enhanced Gaussian Splatting with a Nonlinear Weighted Loss

Authors: Yongfei Guo, Qizhou Huo, Xuan Sun, Yuanhao Gong
Published: 2026-06-06
Categories: cs.CV, cs.GR, cs.MM, eess.IV, math.OC
arXiv: https://arxiv.org/abs/2606.07932v1

3D Gaussian Splatting (3DGS) has become an efficient explicit representation for radiance field reconstruction and real-time novel view synthesis. However, its standard photometric loss treats flat and structure-rich regions similarly, which may limit the recovery of sharp contours and fine details. Edge-Guided Gaussian Splatting (EGGS) improves structure awareness through edge-guided weighting, but mainly relies on first-order gradient responses and linear weighting. In this paper, we propose LEGS, a Laplacian-Enhanced Gaussian Splatting method with a nonlinearly weighted loss. LEGS replaces first-order gradient guidance with second-order Laplacian structural guidance and maps the normalized Laplacian response into pixel-wise weights through nonlinear response-to-weight functions. The proposed loss improves structure-aware Gaussian optimization while keeping the original 3DGS rendering pipeline unchanged. Experiments on the full Tanks\&Temples and Mip-NeRF360 datasets show that LEGS improves peak signal-to-noise ratio (PSNR) by up to 1.68 dB over 3DGS and up to 0.52 dB over EGGS. Incorporating the proposed second-order nonlinear weighting strategy into FastGS and FasterGS further improves PSNR by up to 1.69 dB, demonstrating its effectiveness as a general loss-level extension for Gaussian Splatting pipelines with potential applications in AR/VR, immersive visualization, and real-time 3D content generation.

---

### Pointwise Complexity for Gaussian Fields: Upper Envelopes, Algorithmic Lower Bounds, and Separation

Authors: Yunbei Xu
Published: 2026-06-06
Categories: math.PR, cond-mat.stat-mech, cs.IT, cs.LG, math.ST
arXiv: https://arxiv.org/abs/2606.07931v2

We prove a variance-aware pointwise majorizing-measure theorem for centered Gaussian processes. Classical generic chaining characterizes the scalar quantity $\mathbb E\sup_{x\in T}X_x$; the theorem here gives a simultaneous high-probability envelope for the entire field. For an ambient prior $μ$, the envelope at $x$ is governed by a pointwise Fernique-Talagrand functional \(Φ_μ(x):=\int_0^{4σ(x)}\sqrt{\log\frac{1}{μ(B_d(x,\varepsilon))}}\,d\varepsilon,\) together with the corresponding Gaussian tail term. The theorem provides a reusable field-level refinement of classical generic chaining and a Gaussian-process counterpart of pointwise empirical-process bounds for deep neural networks. We also record a Bayesian algorithmic lower envelope from the interactive Fano/data-processing principle. For a known prior $π$, an observation channel, and a concrete estimator $\widehat t(Y)$, the lower bound is expressed through the exact ghost small-ball mass $\mathbb E_{Y\sim Q}π(B_d(\widehat t(Y),Δ))$, rather than a worst-case covering number. In Gaussian location experiments, comparison decoders convert Bayes location error into lower bounds on decision-aligned Gaussian ranges. We then construct an elementary example separating the usual Fano relaxation, the Bayesian algorithmic lower envelope, the pointwise Gaussian envelope, and the full-class minimax risk. Together, these results show that algorithmic lower bounds provide local-geometric validations of pointwise complexity for fixed estimators in overparameterized ambient classes, precisely in regimes where classical minimax theory becomes either too coarse or oracle-dependent. This separation can also be recast in minimax language as penalty-range information relaxation, highlighting an important question of algorithmic robustness for classical high-dimensional models and regularized algorithms.

---

### Stress-testing medical large language models reveals latent safety pathology beyond benchmark accuracy

Authors: Yuan Shen, Xiaojun Wu, Linghua Yu
Published: 2026-06-06
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.07929v1

Large language models (LLMs) are entering clinical practice based on benchmark accuracy that may fail to detect safety-relevant failure modes. Here we present AI-MASLD, a stress-audit framework that adapts the logic of metabolic stress testing from hepatology to the evaluation of clinical LLMs. Using 240 clinical cases across six narrative perturbation probes, we subjected seven models to double-stress testing and quantified performance through three indices: metabolic index (MI), perturbation flip rate (PFR), and counterfactual fairness index (CFI). Under clean baseline conditions, all models performed uniformly well. Under realistic narrative stress, performance diverged sharply, revealing two distinct stress-response phenotypes. Quantized models exhibited pseudonormalization, in which low flip rates hid functional collapse. Medical supervised fine-tuning systematically degraded logical stability, fairness, and information extraction. An open-weight model matched or exceeded proprietary alternatives on every safety dimension. These findings establish narrative stress auditing as a necessary complement to accuracy-based evaluation.

---

### Barycentric Projections of Optimal Transport Plans on Riemannian Manifolds

Authors: Kisung You
Published: 2026-06-06
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2606.07926v1

Optimal transport couplings are probabilistic objects, while many learning pipelines require deterministic maps. In Euclidean space, barycentric projection converts a coupling into a map by taking conditional expectations, but on a Riemannian manifold curvature and cut loci make this operation nontrivial. We develop a framework for barycentric projections of transport couplings on Riemannian manifolds. The intrinsic projection maps each source point to the conditional Fréchet mean of its destination law and is shown to be the best deterministic representative under squared geodesic loss. The corresponding minimum value is an integrated conditional Fréchet variance, which vanishes exactly for map-induced couplings and therefore defines a conditional-variance Monge defect. We also study a tangential log-exp projection, prove its Euclidean exactness, its compatibility with Brenier-McCann maps in the Monge case, and its interpretation as the first unit Riemannian gradient update for the intrinsic objective. For discrete couplings, both constructions decompose row-wise into weighted Fréchet mean and log-exp problems. Experiments on spherical data, synthetic SPD data, and real EEG covariance matrices support the proposed division of roles: the intrinsic projection is the variational representative, while the tangential projection is a useful local displacement surrogate.

---

### ROSUM-MCTS: Monte Carlo Tree Search-Inspired HDL Code Summarization with Structural Rewards

Authors: Prashanth Vijayaraghavan, Charles Mackin, Luyao Shi, Apoorva Nitsure, Ashutosh Jadhav, David Beymer, Tyler Baldwin, Ehsan Degan, Vandana Mukherjee
Published: 2026-06-06
Categories: cs.CL
arXiv: https://arxiv.org/abs/2606.07925v1

Large language models (LLMs) have shown promise in code summarization, yet their effectiveness for Hardware Description Languages (HDLs) like VHDL and Verilog remains underexplored. We propose ROSUM-MCTS, an LLM-guided approach inspired by Monte Carlo Tree Search (MCTS) that refines summaries through structured exploration and reinforcement-driven optimization. Our method integrates both local and global context via a hierarchical candidate expansion mechanism and optimizes summaries using a composite reward function balancing functional correctness (FC), local content adequacy (LCA), and fluency. We evaluate ROSUM-MCTS on the VHDL-eval and Verilog-eval datasets, demonstrating its consistent outperformance over baseline methods by leveraging structured bottom-up refinement and reinforcement-based optimization. Ablation studies confirm the necessity of both local and global expansion strategies, as well as the importance of balancing FC and LCA for optimal performance. Furthermore, ROSUM-MCTS proves robust against superficial modifications, such as variable renaming, maintaining summary quality where baselines degrade. These results establish ROSUM-MCTS as an effective and robust HDL summarization framework, paving the way for further research into reinforcement-enhanced code summarization.

---

### Decoupling Semantics and Logic: A Training-Free Coarse-to-Fine Pipeline for Video Retrieval-Augmented Generation

Authors: Jiaxin Dai, Zehang Wei, Jiamin Yan, Xiang Xiang
Published: 2026-06-06
Categories: cs.CV, cs.AI, cs.CL, cs.LG, cs.MM
arXiv: https://arxiv.org/abs/2606.07924v1

This paper presents our system description for the 2nd Workshop on Multimodal Augmented Generation via MultimodAl Retrieval (MAGMaR). Addressing the critical challenges of cross-lingual long-video comprehension, strict persona adherence, and zero-hallucination temporal grounding, we propose a fully training-free, two-stage cascaded Video RAG pipeline. Our architecture strategically decouples semantic retrieval from cognitive logical reasoning through a modality-aware division of labor. In the first stage, a high-recall semantic pre-fetching module employs dense retrieval using only high-fidelity visual summaries and global text descriptions, explicitly isolating noisy modalities (e.g., OCR and ASR) to maintain a pristine vector space. In the second stage, an Adaptive, Iterative, and Reasoning-based (A.I.R.) filtering agent, powered by a commercial Large Language Model (LLM), performs fine-grained cognitive reranking. The agent re-incorporates full multimodal contexts to enforce strict logical alignment with user personas, effectively pruning semantically similar but logically irrelevant candidates. Finally, a Prompt Sculpting mechanism constrains the generator to synthesize the distilled subset into strictly formatted JSON responses with exact chunk-level citations. Evaluated on the RAG track, our resource-aware approach shows exceptional precision in both information retrieval and persona-conditioned generation.

---

### Larch: Learned Query Optimization for Semantic Predicates

Authors: Fuheng Zhao, Pawel Liskowski, Zihan Li, Benjamin Han, Puxuan Yu, Varich Boonsanong, Dimitris Tsirogiannis, Anupam Datta
Published: 2026-06-06
Categories: cs.DB, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2606.07923v1

With the advent of Large Language Models (LLMs), many database systems introduced semantic operators that enabled analytical queries over unstructured data (e.g. text, images, videos). Semantic operators typically incur high inference costs and latencies making semantic (AI) SQL queries challenging to apply on large scale datasets. At the same time, their semantic nature leads database engines to treat them as black boxes, making AISQL queries difficult to optimize. In this paper, we introduce Larch, a framework for optimizing the execution of semantic filters in AI SQL queries. Larch was inspired by two key observations: i) the high latency of semantic operators leaves significant room for computationally-heavy runtime optimization techniques, ii) unstructured data are typically accompanied by semantic information in the form of embeddings allowing for efficient semantic comparisons between AI_FILTER prompts and data values. Based on these two key observations, we present two Larch variants: Larch-A2C and Larch-Sel. Larch-A2C encodes arbitrary semantic filters expression tree using an embedding-augmented Gated Graph Neural Network and formulates the filter evaluation order as a Markov decision process. In contrast, Larch-Sel leverages a supervised learning model to predict filter selectivities, subsequently applying dynamic programming to find a near-optimal evaluation order for each input row. Evaluated across diverse real-world datasets and comprehensive synthetic workloads, both Larch variants always outperform existing semantic filter optimization techniques in terms of token usage. Our results demonstrate that Larch is robust across diverse workloads, reducing total token cost overhead by 3x-19x compared to Palimpzest and Quest.

---

### The CIFAR Synthetic Evidence Corpus for Detecting AI-Generated Evidence

Authors: Kelly McConvey, Jalehsadat Mahdavimoghaddam, Nima Jamali, Maksym Taranukhin, Sajad Ebrahimi, Wentao Zhang, Yuntian Deng, Karen Eltis, Maura R. Grossman, Vered Shwartz, Ebrahim Bagheri
Published: 2026-06-06
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.07916v1

The growing ability of generative models to produce realistic documents poses a direct challenge to evidentiary workflows in the justice system and the courts, where decisions increasingly depend on the authenticity of evidence such as receipts, communications, and administrative records. Unlike social media or academic settings, evidentiary documents are often only subtly altered, with small, localized edits that preserve overall plausibility while changing legal meaning. Yet progress on automated detection remains limited, largely due to the absence of suitable training and evaluation data especially suited for the justice system requirements. Existing resources are either focused on photos of human faces or natural scenery or on narrowly scoped academic or social media document types, and do not capture the structure, diversity, or manipulation patterns characteristic of real-world evidentiary data. As a result, current detection systems do not necessarily learn meaningful signals appropriate for the justice system. We introduce the CIFAR Synthetic Evidence Corpus, a dataset designed to enable rigorous evaluation of evidence verification under realistic and controlled conditions. The corpus spans multiple document families and a spectrum of manipulation strategies, from small field-level edits to complete document fabrication, and is constructed using a diverse set of state-of-the-art generative tools. It is organized to systematically vary both manipulation complexity and generation method, while enforcing source-level separation between training and test data to reflect real-world generalization challenges.

---

### EditSR: Enhancing Neural Symbolic Regression via Edit-based Rectification

Authors: Da Li, Xinxin Li, Xingyu Cui, Jin Xu, Juan Zhang, Junping Yin
Published: 2026-06-06
Categories: cs.AI
arXiv: https://arxiv.org/abs/2606.07915v1

Neural symbolic regression models improve inference efficiency by shifting structural search to pretraining, but their one-pass autoregressive decoding is prone to error accumulation, which may lead to generating structurally incorrect expressions, especially in complex expression generation scenarios. Existing rectification strategies can alleviate this issue, but they often depend on restarting global search, thereby weakening the efficiency advantage of neural models, and remain susceptible to error accumulation. In this paper, we propose EditSR, a two-layer framework that combines a neural symbolic regression model in the first layer with an edit-based Rectifier in the second layer to achieve efficient prediction and post-hoc rectification. Instead of restarting the global search, we maintain rectification efficiency by pretraining the Rectifier. Specifically, we formulate the rectification process as a step-by-step state-transition chain starting from an incorrect expression, and develop a state-transition algorithm to construct supervised rectification chains for training the Rectifier. To ensure syntactic validity throughout rectification, each edit action is restricted to a syntactically valid space so that every edited expression remains parseable. In addition, because each edit decision is conditioned on the current state rather than the history, the Rectifier allows errors made in earlier steps to be rectified by subsequent edits, thereby reducing the risk of error accumulation. Extensive experiments and ablation studies show that EditSR substantially improves symbolic structure recovery with limited extra cost, with more pronounced gains on complex expressions, where one-pass autoregressive decoding is more susceptible to error accumulation.

---

### Identifiability and Estimation for Unlabeled Finite Mixtures under Marginal Independence

Authors: Takafumi Kanamori, Yushi Hirose, Shohei Yamamoto
Published: 2026-06-06
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2606.07914v2

We study component recovery and mixing-matrix estimation from unlabeled finite mixtures whose observable distributions share the same latent components but have unknown mixing weights. The main identifying signal is marginal independence: each component is assumed to be independent on at least one coordinate pair, but no labels, clean component samples, or mixing weights are observed. We first prove a structural result for product components: under a subset-rank condition on the spans of the univariate marginals, any independent affine combination of the components must coincide with a single component. We then extend this principle to observable mixtures and show that, under the corresponding subset-rank, full-rank, and no-cancellation conditions, marginally independent affine combinations recover the corresponding latent components. When every component is independent on some coordinate pair, all components are identifiable, and the mixing matrix is recoverable under the stated completion conditions. Finally, we propose a Product-Marginal Maximum Mean Discrepancy (PM-MMD) estimator over affine combinations of the observable mixtures and prove uniform convergence and stability under approximate marginal independence. This framework also separates the empirical roles of the assumptions: irreducibility is, in general, not directly testable from the unlabeled mixtures alone, whereas marginal independence yields a candidate-level diagnostic through held-out PM-MMD. Controlled and flow-cytometry experiments show when marginal independence provides a useful recovery signal. In the reported multi-component comparisons, condition-aware representative selection stabilizes PM-MMD and improves recovery relative to clustering, factorization, and pairwise mixture-proportion baselines using the same unlabeled mixtures.

---

### Identifiability and Estimation for Unlabeled Finite Mixtures under Marginal Independence

Authors: Takafumi Kanamori, Yushi Hirose, Shohei Yamamoto
Published: 2026-06-06
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2606.07914v1

We study component recovery and mixing-matrix estimation from unlabeled finite mixtures whose observable distributions share the same latent components but have unknown mixing weights. The main identifying signal is marginal independence: each component is assumed to be independent on at least one coordinate pair, but no labels, clean component samples, or mixing weights are observed. We first prove a structural result for product components: under linear independence of the univariate marginals, any independent affine combination of the components must coincide with a single component. We then extend this principle to observable mixtures and show that, under full-rank and no-cancellation conditions, marginally independent affine combinations recover the corresponding latent components. When every component is independent on some coordinate pair, all components are identifiable, and the mixing matrix is recoverable under the stated completion conditions. Finally, we propose a Product-Marginal Maximum Mean Discrepancy (PM-MMD) estimator over affine combinations of the observable mixtures and prove uniform convergence and stability under approximate marginal independence. This framework also separates the empirical roles of the assumptions: irreducibility is, in general, not directly testable from the unlabeled mixtures alone, whereas marginal independence yields a candidate-level diagnostic through held-out PM-MMD. Controlled and flow-cytometry experiments show when marginal independence provides a useful recovery signal. In the reported multi-component comparisons, condition-aware representative selection stabilizes PM-MMD and improves recovery relative to clustering, factorization, and pairwise mixture-proportion baselines using the same unlabeled mixtures.

---

### CAAL: Contextual Bandits based Online Hand-Craft Active Learning Strategy Selection

Authors: Shao-An Yin, Jiacong Li, Tianpei Xie, Cecile Levasseur, Wojciech Kowalinski, Nicola Elia
Published: 2026-06-06
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.07910v1

The challenge with active learning algorithms is the uncertainty of the statistical distribution of unlabeled data, making it difficult to choose the best hand-crafted strategy. To address this, we introduced Contextual Adaptive Active Learning (CAAL). In CAAL, each "arm" represents a hand-crafted strategy. Unlike existing frameworks that select strategies based only on feedback from labeled data, we dynamically choose strategies for labeling batches of data using reward prediction with external context information. This general framework allows for customization with domain knowledge to design more effective rewards and context candidates. In addition, we experimentally show that CAAL outperforms the existing baseline adaptive strategy on public datasets using our reward and context design. Our results are consistent regardless of batch size in each iteration.

---

### MemToolAgent: Leveraging Memory for Tool Using Agents Based on Environment and User Feedback

Authors: Suleyman Armagan Er, Danilo Ribeiro, Yogesh Virkar, Surafel Lakew, Adi Kalyanpur, James Gung, Thomas Delteil, Arshit Gupta
Published: 2026-06-06
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2606.07909v2

Modern large language model (LLM) agents can use external tools to help users solve complex tasks. However, for problems that require learning from long-term historical events or from previous agent-environment interactions, LLM agents are required to use memory mechanisms to store and retrieve experiences. While sophisticated memory systems exist for dialogue agents, few studies have empirically examined how to improve agents' tool-using capabilities through past user-agent conversations. We propose MemToolAgent, a framework that improves tool use through memory management. Our approach contains a memory extraction module that processes past experiences into structured memory entries, and a retrieval module that dynamically selects a subset of the stored memory entries. This enables more personalized and accurate responses aligned with user preferences and feedback without requiring LLM fine-tuning. In summary, this work has three main contributions: (1) a unified memory entry format that improves both general-purpose and personalized tool use without LLM fine-tuning, (2) a reflection-based memory extraction that uses environment and user feedback to distill wrong executions into critiques to store, and (3) a retrieval module that chooses how many past experiences to use based on the memory similarity distribution. MemToolAgent achieves 29%, 80%, and 17% relative improvements compared to strong baselines on the WorkBench, NESTFUL, and PEToolBench benchmarks, respectively.

---

### Layer-wise Derivative Controlled Networks Achieve Competitive Accuracy and Gradient Stability Across Data Regimes

Authors: Rowan Martnishn
Published: 2026-06-06
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.07908v2

Derivative-controlled networks based on ChainzRule (CR) combine cubic polynomial layers with a lightweight forward-mode per-layer Jacobian penalty (DREG). In this second paper of a multi-part series, we evaluate the generalization properties of CR across data regimes. We ablate the shape of the DREG coefficient schedule, demonstrating that the optimal annealing range depends on representation noise. On the Pima Diabetes dataset, CR achieves strong low-data performance and maintains a consistent accuracy advantage over baselines from 5\% to 100\% training data, supported by exceptionally stable gradient tail ratios ($\sim$1.01--1.02 vs. 1.07--1.09 for ReLU networks). Extensions to SST-5 show competitive or superior results in both frozen-embedding and BERT fine-tuned regimes, including outperforming prior BERT baselines despite substantially less training data. These results are statistically significant: CR achieves superior accuracy over the strongest published baselines we could identify on both datasets ($p < 0.05$). These results establish that layer-wise derivative control induces a structural inductive bias toward low-frequency, stable representations that generalizes robustly across tabular and NLP domains, data volumes, and representation qualities. The gradient tail ratio serves as a reliable, label-free diagnostic of generalization capability.

---

### Layer-wise Derivative Controlled Networks Achieve Competitive Accuracy and Gradient Stability Across Data Regimes

Authors: Rowan Martnishn
Published: 2026-06-06
Categories: cs.LG
arXiv: https://arxiv.org/abs/2606.07908v1

Derivative-controlled networks based on ChainzRule (CR) combine cubic polynomial layers with a lightweight forward-mode per-layer Jacobian penalty (DREG). In this second paper of a multi-part series, we evaluate the generalization properties of CR across data regimes. We ablate the shape of the DREG coefficient schedule, demonstrating that the optimal annealing range depends on representation noise. On the Pima Diabetes dataset, CR achieves strong low-data performance and maintains a consistent accuracy advantage over baselines from 5\% to 100\% training data, supported by exceptionally stable gradient tail ratios ($\sim$1.01--1.02 vs. 1.07--1.09 for ReLU networks). Extensions to SST-5 show competitive or superior results in both frozen-embedding and BERT fine-tuned regimes, including outperforming prior BERT baselines despite substantially less training data. These results are statistically significant: CR achieves superior accuracy over the strongest published baselines we could identify on both datasets ($p < 0.05$). These results establish that layer-wise derivative control induces a structural inductive bias toward low-frequency, stable representations that generalizes robustly across tabular and NLP domains, data volumes, and representation qualities. The gradient tail ratio serves as a reliable, label-free diagnostic of generalization capability.

---