# Research Papers — 2026-07 · bundle page 189 of 222 > Full-text markdown of papers #9401–9450 of 11075 published in 2026-07, newest first. AI-readable mirrors of the full 11075-paper 2026-07 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-07/page/189 [← Previous bundle](https://aipod.dev/papers/archive/2026-07/page/188.md) [Next bundle →](https://aipod.dev/papers/archive/2026-07/page/190.md) ### On Pairwise Quantile Regression - Statistical Guarantees and Applications Authors: Romain Thérézien, Stephan Clémençon, Fantin Girard, Hamza El-Abdouni Published: 2026-07-05 Categories: stat.ML, cs.AI, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2607.04431v2 Quantile regression provides a powerful tool for summarizing the conditional distribution of a real-valued random variable (r.v.) of interest $Y$ as a function of covariates $Z$ in cases where it shows a large dispersion with high probability, going beyond the situation where standard least square regression is informative/predictive. This article aims to extend this methodology to the pairwise setting, where the variable to be explained is a similarity score between two independent observations (e.g., pixelated ID photos used as input to biometric systems), and the explanatory variables consist of the pair of covariates attached to these observations, such as age or hair color. We establish theoretical guarantees for solutions of this statistical learning problem, considered here as empirical minimizers of a pairwise version of the pinball loss. Leveraging sharp concentration results for $U$-processes, we prove generalization bounds and identify mild conditions under which fast learning rates can be achieved. Confirming the probabilistic analysis, experiments based on simulation data also provide solid empirical evidence of the validity of the methodology promoted here for pairwise quantile regression. Finally, its usefulness from an application perspective is demonstrated by a detailed study aimed at analyzing errors in similarity scoring for facial recognition. --- ### Uncertainty-Aware Abstention in Large Language Models with Provable Alignment Guarantees Authors: Sijin Dong, Hiroyuki Shinnou Published: 2026-07-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2607.04430v1 Large language models (LLMs) are increasingly deployed in question answering (QA) systems, yet they may generate hallucinated or misaligned responses without reliable confidence estimates. Uncertainty quantification (UQ) offers a natural basis for selective answering, where a system answers only when its prediction is deemed reliable and abstains otherwise. However, existing uncertainty scores for LLMs are often heuristic: a threshold chosen on such scores does not, by itself, provide statistical guarantees on the error rate among accepted answers. We propose CIC, a confidence-interval-based calibration framework that converts arbitrary uncertainty scores into risk-controlled selective answering rules. Given a held-out calibration set, CIC evaluates each generated response using an application-specific alignment criterion and associates it with an uncertainty score and a binary error label. For each candidate uncertainty threshold, CIC estimates the acceptance-conditioned error rate and constructs a high-probability upper confidence bound using either Hoeffding-style or Clopper-Pearson confidence intervals. It then selects the largest threshold whose upper bound is below a user-specified risk level $α$, thereby maximizing the answering rate subject to a finite-sample reliability constraint. Under exchangeability, CIC guarantees with probability at least $1-δ$ that the selected threshold, if non-null, controls the error rate among accepted answers at level $α$. We evaluate CIC on both closed-ended and open-ended QA benchmarks across seven LLMs and multiple uncertainty estimators. Experimental results show that CIC consistently achieves valid risk control while retaining strong answering efficiency, providing a practical and statistically grounded mechanism for deploying LLMs in reliability-sensitive QA workflows. --- ### evalci: A Python Library for Statistically Rigorous Comparison of Language Model Evaluations Authors: Shreyas K Chandrahas Published: 2026-07-05 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.04429v1 The dominant practice in language model evaluation is to report a single accuracy number per model and declare the higher one better, without testing whether the gap could plausibly be sampling noise. On benchmarks of a few thousand items, and under temperature sampling where a model can differ from itself run to run by more than the reported gap between models, this practice routinely overstates confidence in headline claims. The statistical machinery to fix this -- confidence intervals, paired significance tests, power analysis, clustered standard errors, multiple-comparison correction -- is well established, but no standard, pip-installable tool packages it in the shape an evaluation actually takes: a per-item results table. We present evalci, a pure-Python library (numpy/scipy/pandas only) that turns a per-item results table into a publication-ready claim -- e.g., "Model A beats Model B, $Δ=3.1$ pts, 95% CI (1.2, 5.0), paired permutation $p=0.002$, $n=1{,}319$" -- in one function call, with adapters for lm-evaluation-harness and HELM output. Every routine is validated against an independent reference (statsmodels, or brute-force exact enumeration) rather than only against itself. As a case study, we re-analyze a public comparison of nine language models' MMLU accuracy and find that 3 of the 8 adjacent leaderboard-rank gaps are not statistically significant after correcting for the 36 pairwise comparisons the ranking implies. evalci is available at https://pypi.org/project/evalci/ (source: https://github.com/Shreyaskc/evalci, DOI: https://doi.org/10.5281/zenodo.21201815) --- ### dOPSD: On-Policy Self-Distillation for Diffusion Language Models Authors: Phuong Tuan Dat, Qi Li, Xinchao Wang Published: 2026-07-05 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2607.04428v1 Diffusion large language models (dLLMs) generate text by iteratively denoising a masked sequence, offering a parallel alternative to autoregressive models, but eliciting strong reasoning through post-training remains difficult: supervised fine-tuning is off-policy and suffers from exposure bias, while reinforcement learning gives only sparse, sequence-level rewards and is hard to apply without tractable sequence likelihoods. On-policy self-distillation (OPSD) offers a promising alternative, using one model as both student and teacher to provide dense, token-level, on-policy supervision, but its effectiveness hinges on giving the teacher privileged information (PI) - typically an instance-specific ground-truth reference unavailable at inference - so the student ends up distilling a weak PI-free consensus policy that yields little improvement on dLLM reasoning. We introduce dOPSD, which instead derives the teacher's privilege directly from the student's own denoising trajectory, evaluating masked positions using later, more-decoded steps of that same trajectory rather than an external label, so the teacher's advantage emerges from the model's own decoding process; on Dream and LLaDA, dOPSD improves both in-domain math reasoning and out-of-domain code generation, outperforming supervised and on-policy baselines. --- ### UI-MOPD: Multi-Platform On-Policy Distillation for Unified GUI Agents Authors: Niu Lian, Tongbo Chen, Zhehao Yu, Chengzhen Duan, Fazhan Liu, Hui Liu, Pei Fu, Jian Luan, Heng Qu, Shu-Tao Xia, Jinpeng Wang Published: 2026-07-05 Categories: cs.CL, cs.AI, cs.CV, cs.LG, cs.MM arXiv: https://arxiv.org/abs/2607.04425v2 Recent advances in multimodal foundation models and agent systems have driven GUI agents from single-platform task execution toward cross-platform interaction. However, unified multi-platform GUI learning remains challenging: high-quality cross-platform trajectories remain scarce, while platforms share transferable capabilities but differ in action semantics and interaction conventions. Naively mixing supervision or merging specialized models can blur native behaviors and produce imbalanced performance. To address these challenges, we construct Uni-GUI, a high-quality dataset containing nearly 10K executable cross-platform interaction trajectories collected through a unified desktop-mobile harness. Building on Uni-GUI, we propose UI-MOPD, the first framework to introduce multi-teacher on-policy distillation (MOPD) into unified multi-platform GUI agent training. UI-MOPD trains a shared student on its own rollouts and dynamically routes each rollout to the corresponding platform-specialized teacher. At student-visited states, teacher guidance serves as a platform-conditioned behavioral anchor, enabling the integration of complementary desktop and mobile expertise without averaging their distinct interaction conventions. On OSWorld and MobileWorld, UI-MOPD achieves task success rates of 38.2% and 12.0%, respectively, outperforming parameter-matched integration strategies while preserving general GUI grounding. These results demonstrate that multi-teacher on-policy distillation provides an effective approach to building unified cross-platform GUI agents. Project page: https://elispectre.github.io/UI-MOPD/. --- ### UI-MOPD: Multi-Platform On-Policy Distillation for Continual GUI Agent Learning Authors: Niu Lian, Alan Chen, Zhehao Yu, Chengzhen Duan, Fazhan Liu, Hui Liu, Pei Fu, Jian Luan, Yaowei Wang, Shu-Tao Xia, Jinpeng Wang Published: 2026-07-05 Categories: cs.CL, cs.AI, cs.CV, cs.LG, cs.MM arXiv: https://arxiv.org/abs/2607.04425v1 Recent advances in multimodal foundation models and agent systems have driven GUI agents from single-platform task execution toward cross-platform interaction. However, building multi-platform GUI agents remains challenging. On one hand, high-quality and executable cross-platform interaction trajectories are still scarce, and existing data often suffer from limited platform coverage. On the other hand, different platforms exhibit distinct interaction conventions, making joint or continual training prone to behavioral pattern mixing, platform-specific capability degradation, and catastrophic forgetting. To address these challenges, we construct Uni-GUI, a high-quality cross-platform GUI interaction dataset, and propose UI-MOPD, the first method that incorporates multi-teacher on-policy distillation into continual learning for GUI agents. UI-MOPD dynamically selects a platform-specific teacher according to the current environment and transfers platform-specific behavioral priors to a shared policy through platform-conditioned distillation, enabling adaptation to new platforms while preserving capabilities on existing ones. Experiments on OSWorld and MobileWorld show that UI-MOPD achieves task success rates of 38.2% and 12.0%, respectively, demonstrating its effectiveness in balancing cross-platform capability retention and new-platform adaptation. Project page: https://elispectre.github.io/UI-MOPD/. --- ### Transferability Between Understanding and Generation in Unified Multimodal Models Authors: Jiwon Kang, Heeji Yoon, Jaewoo Jung, Jaewon Min, Minkyeong Jeon, Biyeon Hwang, Sangwon Jung, Seungryong Kim Published: 2026-07-05 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2607.04423v2 Unified Multimodal Models (UMMs) integrate image understanding and generation within a single architecture, yet how the two tasks interact remains understudied. We investigate $\boldsymbol{\mathsf{transferability}}$ in UMMs: whether training a capability on one task improves the same capability on the other without explicit supervision. Through controlled experiments, we empirically find that transferability depends on architecture-models with fully shared transformer backbone and a unified visual encoder exhibit consistent cross-task transfer, while loosely coupled designs show little or none. Leveraging this transferability, we propose a practical training strategy. The most straightforward way to improve a target generative capability (e.g., counting) is to fine-tune generation directly, but this can degrade visual quality due to distribution shift. Instead, we train the corresponding understanding task and let it transfer into generation, which improves capability-specific generative performance while minimizing distribution shift. We validate this across three capabilities-counting, spatial relation, and text recognition/generation-showing that cross-task transferability can be systematically exploited in UMMs. --- ### Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Authors: Siyu Ding, Mingchuan Ma, Jiabo Tong, Xingrun Xing, Ziming Wang, Guoqi Li Published: 2026-07-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2607.04422v2 Recent NVFP4 pretraining work has primarily optimized Transformer linear projections, leaving persistent optimizer states, optimizer computation, and low-precision attention forward--backward paths less explored. We present \textbf{Full-Stack FP4}, a modular NVFP4 framework with separate recipes for projections, AdamW states, Root/Muon computation, and attention. \textbf{LoRA-SVD} protects a compact projection subspace in BF16 while retaining full-shape NVFP4 computation, reducing the linear-only loss gap from \textbf{1.40\%} to \textbf{0.61\%}. An ordered square-root, tile-mean, and Hadamard pipeline enables stable NVFP4 AdamW momentum storage; shape-dependent coefficients and clipping stabilize direct NVFP4 Root iterations; and mixed-precision attention retains softmax-sensitive operations in BF16. On 3B pretraining with 64B tokens, BF16 and Full-Stack FP4 reach losses of \textbf{2.267} and \textbf{2.286}, a \textbf{0.838\%} gap. Their average zero-shot perplexities are 26.675 and 26.665, respectively, with Full-Stack FP4 averaging 0.10 percentage points lower in accuracy. Native four-block measurements on one RTX 5090 show 2.50--2.83$\times$ Root speedups over optimized BF16 and 37.9--42.5\% lower AdamW peak memory. --- ### Full-Stack FP4: Stable LLM Pretraining with Quantized Projections, Optimizers, and Attention Authors: Siyu Ding, Mingchuan Ma, Jiabo Tong, Xingrun Xing, Ziming Wang, Guoqi Li Published: 2026-07-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2607.04422v1 Recent NVFP4 pretraining methods mainly target transformer linear layers, leaving optimizer states, optimizer arithmetic and attention underexplored in 4-bit pipelines. This critical gap blocks stable full-stack 4-bit pretraining, as the three core modules exhibit unique numerical failure patterns: linear layers hit hard quantization noise limits with dimension-propagated error amplification; AdamW second moments are heavy-tailed non-negative values fragile to low-precision denominators; attention carries error-prone computation paths demanding strict forward-backward quantization consistency. We propose Full-Stack FP4, the first complete NVFP4 pretraining framework resolving all three stability bottlenecks via module-wise precision strategies. For linear projections, LoRA-SVD lightweight decomposition suppresses quantization noise and breaks the direct-quantization error ceiling, shrinking the linear-only loss gap from 1.40% to 0.61%. For optimizers, we design AdamW second-moment transformation for robust NVFP4 storage and fully support native NVFP4 Newton-Schulz iterations for the Root (Muon) optimizer. For attention, a mixed-precision scheme quantizes Q/K/V and backward dS while guarding vulnerable paths in BF16, paired with unified tensor reuse to sustain forward-backward alignment. We further analyze fast error accumulation in naive low-bit matrix multiplication and the extreme sensitivity of PV / dOV^T attention branches. All modules are plug-and-play with cumulative stability and efficiency improvements. Our 3B/64B-token pretraining validates near-BF16 performance with merely 1.47% loss gap, verifying feasible stable end-to-end NVFP4 LLM pretraining. --- ### Agent Step Value: Auditing Evaluator-Channel Reversals in Black-Box Agent Traces Authors: Andrew Zhang, Chengzhan Li Published: 2026-07-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.04419v4 Pooling, substituting, or reusing evaluator-derived step rewards assumes that their direction survives a change of evaluation channel. The same frozen transition can violate that assumption. Process rewards vary agent states, while evaluator audits vary scoring configurations; neither first difference isolates their interaction. We define Agent Step Value (ASV) as a channel-indexed target-margin gain and identify the state-by-channel interaction on complete matched faces. Across frozen PubMed question-answering transitions, direct scoring yields a positive mean ASV, while the generated-view channel yields a negative mean. Two matched replay waves reproduce this reversal, and cross-channel sign disagreement exceeds same-channel retry disagreement by 48.0 percentage points. Matched retrieval faces localize the reversal to the generated-view coordinate and trace its direction across a readout-and-stack bridge. A source-only generation contract restores the positive mean direction on artifact-bearing retrievals and removes parser-detected substantive support claims from artifact-free before-state views. ASV turns channel sensitivity into an identified measurement problem that can be localized and tested by intervention before step rewards are reused. --- ### Environmental Drivers of Respiratory Disease: A District Level Analysis Authors: Rahim Iqbal, Asfi Ahamed, Izzath Nisfer, Shazan Shaheed, Muhammadu Ilham, Nathali Athukorala, Madara Mendis, Nisansa de Silva, Sandareka Wickramanayake Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04416v2 Sri Lanka has experienced a decade of progressive forest degradation and rising atmospheric pollution, yet district-level respiratory admissions have paradoxically declined, pointing to the confounding role of healthcare access. This study addresses that gap by constructing an 11-year (2014-2024) panel dataset across all 25 administrative districts, integrating satellite-derived vegetation indices, fire radiative power, pollutant concentrations (particulate matter (PM2.5), nitrogen dioxide (NO2), sulfur dioxide (SO2)), carbon flux metrics and population-normalized respiratory admission rates. Two temporally validated XGBoost models were created for annual district-level respiratory rate (R^2 = 0.937) and monthly PM2.5 concentration (R^2 = 0.976) with generalization validated in 21 out of 25 districts (Mean Absolute Percentage Error (MAPE) <= 20%). Shapley Additive Explanations (SHAP) analysis established that cumulative air quality burden is the overwhelming driver of respiratory rate variance (80.1%), ahead of forest degradation (15.6%) and fire activity (4.3%). The Forest-Air-Health (FAH) Risk Index used these SHAP-derived weights to find the districts with the highest risk: Colombo (FAH = 0.802), Gampaha (0.708), and Kalutara (0.682). These findings present the inaugural evidence-based, district-level framework correlating environmental degradation with respiratory health in Sri Lanka, establishing a quantitative basis for focused public health and environmental policy. --- ### LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RL Authors: Yujin Kim, Namgyu Ho, Sangmin Hwang, Joonkee Kim, Yongjin Yang, Sangmin Bae, Seungone Kim, Jaehun Jung, Se-Young Yun, Hwanjun Song Published: 2026-07-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.04412v2 Reinforcement learning (RL) for non-verifiable instruction following increasingly relies on LLM judges with prompt-specific rubrics as reward signals. While recent methods adapt these rubrics to the evolving policy during training, the training prompts themselves remain static, drawn from fixed corpora. This static approach often results in a critical misalignment between prompt difficulty and policy capability, leaving the judge unable to recover a discriminative reward signal when prompts fail to elicit quality variance among rollouts. To address this misalignment, we introduce LLM-as-a-Tutor, a framework that extends the LLM's role from judge to tutor: a single model serves as an examiner that pairwise-compares policy rollouts to detect non-challenging prompts, and as a generator that appends atomic constraints to them. This append-only design monotonically raises difficulty in step with the policy's capability, producing a self-calibrating training signal without external difficulty schedules. On three complex instruction-following benchmarks, our method consistently outperforms both policy-unaware baselines and prior policy-adaptive methods that adapt rubrics or rewrite prompts, suggesting prompt adaptation as a missing axis of policy-awareness in non-verifiable RL. --- ### LLM-as-a-Tutor: Policy-Aware Prompt Adaptation for Non-Verifiable RL Authors: Yujin Kim, Namgyu Ho, Sangmin Hwang, Joonkee Kim, Yongjin Yang, Sangmin Bae, Seungone Kim, Jaehun Jung, Se-Young Yun, Hwanjun Song Published: 2026-07-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.04412v1 Reinforcement learning (RL) for non-verifiable instruction following increasingly relies on LLM judges with prompt-specific rubrics as reward signals. While recent methods adapt these rubrics to the evolving policy during training, the training prompts themselves remain static, drawn from fixed corpora. This static approach often results in a critical misalignment between prompt difficulty and policy capability, leaving the judge unable to recover a discriminative reward signal when prompts fail to elicit quality variance among rollouts. To address this misalignment, we introduce LLM-as-a-Tutor, a framework that extends the LLM's role from judge to tutor: a single model serves as an examiner that pairwise-compares policy rollouts to detect non-challenging prompts, and as a generator that appends atomic constraints to them. This append-only design monotonically raises difficulty in step with the policy's capability, producing a self-calibrating training signal without external difficulty schedules. On three complex instruction-following benchmarks, our method consistently outperforms both policy-unaware baselines and prior policy-adaptive methods that adapt rubrics or rewrite prompts, suggesting prompt adaptation as a missing axis of policy-awareness in non-verifiable RL. --- ### AI Wizards at EXIST 2026: Hierarchical Soft-Label Learning for Multimodal Sexism Identification in Memes Authors: Matteo Fasulo, Antonio Gravina, Luca Tedeschini, Luca Babboni Published: 2026-07-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2607.04410v1 We present the AI Wizards submission to EXIST 2026 for multimodal sexism identification in memes. The task is composed of three, increasingly harder subtasks. We model them hierarchically as conditional soft-label prediction over empirical annotator distributions. Our system maps fixed Gemini Embedding 2 vision-language representations through a lightweight Gated MLP trained with KL divergence and homoscedastic uncertainty weighting. Our submissions ranked first on Task 2.3 and fourth on Tasks 2.1 and 2.2 on the official Soft-Soft leaderboards. The code is available at https://github.com/NLP-AI-Wizards/EXIST-2026 --- ### Learning Task-Sufficient World Models by Synergizing Agentic Exploration and Structured Modeling Authors: Fan Feng, Yujia Zheng, Minghao Fu, Yongqiang Chen, Guangyi Chen, Kevin Murphy, Biwei Huang, Kun Zhang Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04409v1 Learning and planning in imagination using world models provides an effective paradigm for training agents for decision-making. However, existing approaches often rely on high-dimensional latent spaces or generic visual embeddings that retain many factors irrelevant to control, limiting efficiency and generalization across tasks. To this end, we study how agents can learn world models with representations that are task-specific, minimal, and sufficient for decision-making. We achieve this via a closed-loop synergy between the agent and the world model, in which structured world-model learning distills task-sufficient representations from informative interaction data. On the agent side, agents actively probe the environment to collect informative trajectories that expose task-relevant latent factors, guided by an adaptive curriculum. On the world-model side, we learn structured representations over observations to distill compact, task-sufficient latent states from the collected interaction data. This synergy enables the empirical recovery of task-sufficient latent representations that capture all control-relevant factors. Leveraging these representations, the resulting policies achieve improved sample efficiency and generalization, including generalization across skills, object-skill compositions, and previously unseen tasks on standard continuous-control and robotic-manipulation benchmarks. --- ### Quadrature-Aware Complex-Linear Neural Operator for Boundary-to-Field Prediction in Resonant Acoustics Authors: Muhammad Idrees Khan, Hua-Dong Yao Published: 2026-07-05 Categories: physics.flu-dyn, cs.LG arXiv: https://arxiv.org/abs/2607.04407v1 Repeated prediction of acoustic fields from spatially distributed boundary excitation is computationally expensive when each source realization requires a new wave simulation. This work introduces a quadrature-aware complex-linear boundary operator (CLBO) that maps complex normal velocity on a vibrating surface to complex pressure at receiver locations. The model couples learned source and receiver basis functions through an explicit complex surface-quadrature contraction, so the boundary excitation enters linearly by construction. This preserves complex superposition, homogeneity, and zero response to zero excitation, while representing the source through coordinates, normals, and quadrature weights rather than a fixed flattened input vector. Reference data were generated using a verified three-dimensional multiple-relaxation-time (MRT) lattice Boltzmann solver and stored in a solver-agnostic boundary-to-field format. CLBO was compared with a fixed-sensor complex DeepONet under matched case splits and optimization settings, with additional tests of structural consistency, receiver-coordinate interpolation, source discretization, source-family holdout, label efficiency, physics-informed ablations, unseen source mixtures, and computational cost. Across five training seeds, CLBO achieved a mean complex relative field error of 0.184 +/- 0.00771, compared with 0.367 +/- 0.00742 for DeepONet. Its measured source-superposition error was 1.31 x 10^-7, and its mean error on newly simulated mixed-source cases was 0.237, compared with 0.415 for DeepONet. Inference was 1.83 x 10^4 faster than the reference calculation for the reported query size. These results show that enforcing the known complex-linear boundary-to-field structure improves physical consistency and generalization under distributed acoustic excitation. --- ### MambaRefine-CD: MambaVision with Region-Boundary Temporal Refinement Authors: Dineth Perera, Thaariq Firdous, Oshadha Samarakoon, Roshan Godaliyadda, Parakrama Ekanayake, Vijitha Herath Published: 2026-07-05 Categories: eess.IV, cs.CV arXiv: https://arxiv.org/abs/2607.04403v1 Binary change detection in remote sensing requires both complete changed-region localization and accurate boundary delineation. We present MambaRefine-CD, a region-boundary temporal refinement framework built on a shared MambaVision encoder. The proposed D-RBI module constructs temporal evidence from paired features, absolute differences, and signed differences, then separates it into region and Sobel-conditioned boundary streams. Region features are enhanced with CRAM-lite and decoded by an adaptive receptive-field FPN, while the finest boundary stream guides a bounded residual refinement of the coarse prediction. Experiments on DSIFN-CD and WHU-CD show strong changed-class F1 and IoU under verified evaluation settings, and ablations support the contribution of signed temporal evidence and the full region-boundary refinement pipeline. --- ### The Good, the Bad, and the Brittle: Benchmarking Robustness and Generalisation of Histopathology Foundation Models Authors: Dhyey Yajnik, Amina Asif, Fayyaz Minhas Published: 2026-07-05 Categories: cs.CV, cs.LG arXiv: https://arxiv.org/abs/2607.04401v1 How robust and generalisable are pathology foundation models and have their scaling limites been reached? We benchmarked twelve pathology foundation models (PFMs) and ResNet baselines using our Robustness Evaluation and Enhancement Toolbox (REET) across eleven clinically realistic perturbations and a dissimilarity-driven Non-Redundant K-fold validation (NR-Kfold) protocol. We introduce a Perturbation Performance Index (PPI) to summarise accuracy trends under controlled perturbation sweeps and analyse robustness scaling with parameter count. We show that PFMs consistently outperform CNNs in both robustness and domain generalisation, yet model scaling shows diminishing returns: mid-sized models such (UNI2/Virchow-2 etc.) achieve comparable or greater resilience than larger systems. NR-Kfold analysis further reveals systematic accuracy loss and increased variability when training-test similarity is broken, underscoring the need for explicit distribution-shift evaluation. These findings suggest that the next generation of pathology foundation models must prioritise data quality, multimodality information and domain alignment over parameter count to achieve genuine clinical reliability. --- ### NKI-Agent: Domain-Specific Fine-Tuning and Agentic Tool Use for Neuron Kernel Generation Authors: Junjie Tang, Jun Huan, Hao Zhou, Yuhao Zhang, Lin Wang Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04395v1 Recent agentic approaches to LLM-based kernel generation have achieved impressive results on CUDA. For emerging AI accelerators such as AWS Trainium and Inferentia, automated kernel generation and optimization remain largely unaddressed. Writing kernels for these chips via the Neuron Kernel Interface (NKI) is particularly challenging: developers must navigate a multi-engine architecture, tile-based programming, and explicit data movement across multi-level memory hierarchy. Moreover, no publicly-available training data, benchmarks, or tool-augmented agents exist for this domain. We introduce NKI-Agent, the first system combining domain-specific supervised fine-tuning (SFT) with a compile-verify-fix agent loop for NKI kernel generation. We adapt the existing CUDA-Agent framework to Neuron hardware, curate 6,000 NKI kernel generation tasks for training, and construct NKIBench, a 250-task benchmark across three difficulty levels. Evaluated on real Trn1 hardware, NKI-Agent with Claude Opus 4.8 and a rank-aware system prompt achieves a 77.3% pass rate on the 150-task NKIBench. We show that tool use is critical: Opus 4.8 scores 6% in single-shot mode without agent tools. On a 60-task subset, we show that an SFT-trained Qwen3-Coder-30B-A3B achieves 25.0% pass rate at 1/100th the cost, outperforming Claude Sonnet 4 (15.0%). We also report that Group Relative Policy Optimization (GRPO) with binary compilation reward fails to improve over SFT, providing guidance on reward design for RL-based kernel generation. --- ### MechMath Agent Team: LLM Driven Agents for Mathematical Research Authors: Yichuan Cao, Ruichen Qiu, Junqi Liu, Jiaqi Wang, Dakai Guo, Ruyong Feng, Lihong Zhi, Xiao-Shan Gao Published: 2026-07-05 Categories: cs.AI, cs.SC arXiv: https://arxiv.org/abs/2607.04394v1 AI reasoning has become a central focus in contemporary artificial intelligence, largely driven by the success of large language models. However, mathematical research, which is characterized by non-linear derivation paths, rigorous logical requirements, and protracted exploration cycles, poses severe challenges for existing reasoning systems. To overcome these limitations, we present the MechMath Agent Team (MMAT), which is a large language model driven agent designed to serve as a co-pilot throughout the full cycle of mathematical research. We design a tripartite Harness Architecture that decouples system responsibilities into Control, Execution, and Augmentation planes, thereby reconciling rigorous logical control with the agility demanded by open-ended research. Building upon this framework, we instantiate three specialized agents: a Knowledge Base Manager, a Natural Language Prover, and a Formal Language Prover, all operating in a closed loop to produce formally certified mathematical proofs. We evaluate MMAT on open problems in Number Theory, Algebraic Complexity Theory, Differential Algebra, Operator Algebra, and Inequalities. Across a two-month deployment, 11 problems have been solved, demonstrating its capacity to act as a co-pilot throughout the entire research cycle. The contributions are threefold: a general decoupled Harness Architecture for multi-agent mathematical reasoning, its concrete instantiation in the MMAT system, and empirical validation on a diverse suite of open problems. --- ### Memory-Orchestrated Semantic System (MOSS): An Auditable Agentic Memory Architecture Authors: Serge Lacasse, Jérémie Hatier, Alex Baker Published: 2026-07-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2607.04391v1 Long-term memory remains a structural weakness of AI agents. The dominant approach, retrieval-augmented generation (RAG), relies on embedding-based similarity search, which is opaque by construction, difficult to audit, and bounded by the theoretical limits of vector representations. We present the Memory-Orchestrated Semantic System (MOSS), an agentic memory architecture in which the agent drives retrieval over a structured relational database. MOSS is model-agnostic, storage-agnostic, and API-agnostic: it runs on any relational engine, connects to any LLM provider (or to deterministic non-LLM processes), and deploys on any infrastructure, local or cloud. Its retrieval execution is symbolic and reproducible (once a query is formulated, no LLM participates in the retrieval loop) and every step of the system, from indexing to answer formulation, is logged and inspectable, making MOSS auditable by construction. Rather than imposing an external ontology, MOSS derives its conceptual vocabulary from the corpus itself. We report on a longitudinal deployment unique in the agentic-memory literature: a year of continuous production over an individual scholar's working corpus--a conversational corpus reaching back to October 2024 (some 44 million tokens, retroactively indexed) comprising 110,183 segments, alongside 163,494 catalogued documents, 569 inductively derived concepts, 322,662 concept annotations, and eleven metadata graphs totaling approximately five million relations--across four successive infrastructure generations. While the present case is that of a single researcher, the architecture is in no way specific to one person: it serves a team, an institution, or any entity that accumulates knowledge over time. We argue that auditable, sovereign, structurally unbounded memory is a precondition for AI agents intended to accompany a person or an organization over years rather than sessions. --- ### Decentralized Aggregation of LLM Predictions via Wagering Mechanisms Authors: Yuhong Luo, David M. Pennock, Xintong Wang Published: 2026-07-05 Categories: cs.AI, cs.GT arXiv: https://arxiv.org/abs/2607.04389v1 It is increasingly common to aggregate predictions from multiple LLMs, each with domain expertise or access to private tools and data, to improve collective prediction performance. In decentralized settings, aggregation weights need to be determined without access to models' private information and should remain robust to strategic reporting. We propose a family of advantage-aligned wagering mechanisms for LLM aggregation (WALLA), in which each model reports a prediction and a learned wager, and predictions are aggregated using wagers as weights. WALLA introduces a leave-one-out baseline into the net payout function, yielding three desirable properties: (1) dominant-strategy incentive compatibility of prediction under arbitrary belief structure, (2) advantage--wager alignment, where the optimal wager is proportional to the model's expected score advantage, and (3) prediction-agnostic wager optimization, enabling decentralized learning of wager policies without requiring optimal predictions. We further instantiate two mechanism variants that trade off normality and no-arbitrage while maintaining a bounded worst-case deficit for the mechanism. Experiments on question-answering and forecasting benchmarks across heterogeneous models and private-information settings show that WALLA matches centralized aggregation methods in predictive performance, while simultaneously achieving decentralized learning, advantage-aligned aggregation weights, uncertainty awareness, and incentive-compatible prediction. --- ### Auto-AEG: Scalable Data Construction for Open-Vocabulary Audio Event Grounding Authors: Zihan Zhang, Xize Cheng, Wenhao Yan, Tong Zhang, Dongjie Fu, Boyun Zhang, Yongbo He, Tao Jin Published: 2026-07-05 Categories: cs.SD, cs.AI arXiv: https://arxiv.org/abs/2607.04383v4 Large Audio-Language Models (LALMs) reason fluently about sound yet struggle to localize precisely when events occur, while classical Sound Event Detection attains frame-level precision only over a closed label set. At the intersection of these paradigms lies the task of Open-Vocabulary Audio Event Grounding: predicting all time intervals of a target sound event described by an arbitrary natural language query. Progress is bottlenecked by data scarcity: no large-scale resource provides open-vocabulary onset/offset supervision, and manual temporal annotation is prohibitively expensive. To address this, we introduce Auto-AEG, a scalable pipeline that constructs such supervision by automatic data construction and model fine-tuning. It pairs programmatically synthesized clips, which carry placement-exact ground-truth intervals for supervised cold-start, with multi-model pseudo-labels on real-world audio that supply the reward signal for reinforcement learning. Training with this pipeline yields large temporal-localization gains (+73.9% and +23.1% mIoU over zero-shot) on AEGBench, an independent difficulty-stratified benchmark we release, and these gains generalize to held-out SED and other audio grounding benchmarks. Our results show that automatically constructed data, coupled with interval-aware reward design, provides an effective data-side route to expanding the temporal localization capability of LALMs. AEGBench: https://huggingface.co/datasets/zihan-audio/AEGBench --- ### Event Detection in Videos: A Framework for the Development of New Methods Authors: Anastasia Zakharova, Thierry Bouwmans, Anthony Cioppa, Adrien Deliège, Antonio Greco, Anaïs Halin, Kamil Jeziorek, Meghna Kapoor, Tomasz Kryjak, Islam Osman, Sébastien Piérard, Carlo Sansone et al. Published: 2026-07-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.04372v1 Event detection tasks in videos, the most important aspect of video surveillance, aim to detect events either at the pixel-level, frame-level, or clip-level. Plenty of methods intended for event detection in different environments, for various applications, and within different acquisition techniques were introduced. Naturally, the attempts were made as well to classify these algorithms in terms of detection of performance or in terms of real-time abilities. Nevertheless, the lack of a large-scale dataset as well as rigorous performance evaluation methods have biased such comparisons as well as the development of the methods. Given the diversity of existing approaches, we believe it is essential for researchers to position their work within such a rich landscape. Thus, we propose a rigorous framework for developing new methods in event detection for videos. Specifically, this framework is based on three main pillars: datasets, performance evaluation, and scenarios for deploying methods. --- ### Nemotron-Labs-3-Puzzle-75B-A9B: Compressing Hybrid MoE LLMs Authors: Akhiad Bercovich, Talor Abramovich, Daniel Afrimi, Shay Aharon, Nir Ailon, Vladimir Anisimov, Omer Ullman Argov, Maor Ashkenazi, Tomer Asida, Nave Assaf, Tomer Bar Natan, Alexander Bukharin et al. Published: 2026-07-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.04371v2 We present Nemotron-Labs-3-Puzzle-75B-A9B, a compressed variant of Nemotron-3-Super optimized for interactive deployment. We designed the model to maximize server throughput under high user throughput constraints. In interactive serving workloads on a single 8xB200 node, Puzzle-75B-A9B achieves approximately 2x higher server throughput than Nemotron-3-Super at matched user throughput constraints. In ultra-long-context deployment on a single H100 GPU, the compressed model increases 1M-token concurrency from 1 request to 8 requests. Puzzle-75B-A9B is constructed using a multi-stage pipeline that combines the Iterative Puzzle compression framework with knowledge distillation, reinforcement learning, quantization, and a Multi-Token Prediction head. The compression process jointly optimizes heterogeneous MoE pruning, active parameter budget, and Mamba pruning to improve inference efficiency while preserving model quality. We evaluate Puzzle-75B-A9B on a broad suite of reasoning, coding, multilingual, long-context, and agentic benchmarks. Despite substantial compression, the model retains strong downstream accuracy relative to the parent model across a wide range of tasks. These results demonstrate that large hybrid MoE models can be substantially optimized for deployment efficiency while maintaining strong downstream capability. Our model is publicly available on Hugging Face. --- ### RL Forgets! Towards Continual Policy Optimization Authors: Mao-Lin Luo, Zhe-Xu Wang, Zi-Hao Zhou, Bo Ye, Jian Zhao, Min-Ling Zhang, Tong Wei Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04364v3 Continual post-training is becoming a central paradigm for adapting vision-language models to evolving tasks. Recent studies report that reinforcement learning is less prone to forgetting than supervised fine-tuning, motivating the view that RL is inherently resistant to forgetting. However, this view remains insufficiently validated, as existing evidence is largely drawn from outdated or homogeneous benchmarks. We revisit this assumption by introducing MRCL, a Multimodal Reasoning Continual Learning benchmark built from recent and diverse multimodal reasoning tasks. Experiments on MRCL show that standard reinforcement learning still suffers from catastrophic forgetting during continual post-training. We trace the failure to an objective mismatch. The KL regularization used in common policy optimization methods is evaluated on current-task data, whereas forgetting is caused by behavioral drift on prior-task distributions. To address this problem, we propose Continual Policy Optimization (CPO), a replay-free method grounded in a prior-task behavioral KL objective. CPO derives a local Fisher surrogate from the historical KL objective and uses parameter movement as a gradient-free proxy for Fisher sensitivity, enabling sparse regularization with negligible additional overhead. Experiments on three model scales and comparisons with multiple RL baselines show that CPO consistently reduces forgetting while maintaining effective adaptation and preserving broader pretrained capabilities. The implementation code is available at https://github.com/MaolinLuo/CPO. --- ### RL Forgets! Towards Continual Policy Optimization Authors: Mao-Lin Luo, Zhe-Xu Wang, Zi-Hao Zhou, Bo Ye, Jian Zhao, Min-Ling Zhang, Tong Wei Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04364v2 Continual post-training is becoming a central paradigm for adapting vision-language models to evolving tasks. Recent work has increasingly favored reinforcement learning over supervised fine-tuning, driven by the belief that reinforcement learning is inherently less prone to forgetting. However, the belief remains insufficiently validated, as existing evidence is largely drawn from outdated or homogeneous benchmarks. We revisit this assumption under recent and diverse multimodal reasoning tasks. To this end, we introduce MRCL, a Multimodal Reasoning Continual Learning benchmark. Experiments on MRCL show that standard reinforcement learning still suffers from severe catastrophic forgetting during continual post-training. We trace this failure to an objective mismatch: the KL regularization used in common policy optimization methods is evaluated on current-task data, whereas forgetting is caused by behavioral drift on prior-task distributions. To address this problem, we propose Continual Policy Optimization (CPO), a replay-free framework grounded in a prior-task behavioral KL objective. CPO relaxes the intractable historical KL constraint into sparse parameter-movement regularization, limiting policy drift without storing old data. Extensive experiments across multiple model scales show that CPO consistently reduces forgetting while preserving, and in some cases improving, pretrained model capabilities. On Qwen3-VL-8B, CPO reduces forgetting by 13.7% and improves pretrained capability by 7.0%. The implementation code is available at https://github.com/MaolinLuo/CPO. --- ### Optimal Mixture-of-Experts Model Averaging for Conditional Generative Models Authors: Shijin Gong, Baihua He, Xinyu Zhang Published: 2026-07-05 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2607.04360v1 Conditional generative models have emerged as powerful tools for sampling from target conditional distributions, driving substantial advances across a wide range of scientific and applied domains. As these models proliferate, practitioners often face multiple plausible generators whose performance can vary with the task, data, or input condition. We propose an optimal model averaging framework for conditional generative models, allowing candidate generators to be combined even when they are accessible only through conditional samples without tractable densities. Specifically, we use a sample-based maximum mean discrepancy between conditional distributions, which first leads to a static model averaging method, StaticMA, assigning fixed weights to different candidates. In addition, we develop MoEMA (mixture-of-experts model averaging), an input-adaptive method that parameterizes covariate-dependent weights through a softmax neural-network gate. We establish in-sample and out-of-sample asymptotic optimality for the proposed methods, together with consistency of the estimated adaptive weight function under regularity conditions. The framework applies directly to Euclidean responses and extends to unstructured data by combining our formulation with fixed representation maps. Across a broad set of simulations and real-data studies spanning tabular, image, and text modalities, MoEMA generally improves over competing baselines, demonstrating the effectiveness of our proposed methods. --- ### How Many Initial Points Does Bayesian Optimization Need? Authors: Mujin Cheon, James Odgers, Dong-Yeun Koh, Calvin Tsay Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04356v1 Bayesian Optimization (BO) generally begins with an initialization phase: a batch of $n_0$ uninformed evaluations. The choice of $n_0$ remains largely heuristic, and we empirically observe that the total cost (random initial points plus BO iterations needed to find the global optimum) is U-shaped in $n_0$, i.e., a practitioner wastes resources by selecting either too low or too high a value of $n_0$. We find this tradeoff persists across MLE, Bayesian MCMC, and exact GP hyperparameters, as well as across acquisition functions. Toward the latter, Thompson Sampling appears an exception, with both total cost and simple regret essentially $n_0$-agnostic, though higher in our experiments. We attribute this U-shape to the known boundary issue of variance-driven BO: BO burns early budget on corners of the hypercube before turning inward. We demonstrate this effect using a 3D BO trajectory where the exact hyperparameters are known. We conclude with practical recommendations: use multi-step lookahead BO where possible; otherwise use Thompson Sampling when $n_0$ cannot be tuned, and a generously large $n_0$ when it can. --- ### HASSL: Hierarchy-Aware Self-Supervised Learning Framework for Single Cell Microscopy Authors: Julius Riel, Vishwa Mohan Singh, Sai Anirudh Aryasomayajula, Anuun Chinbat, Hannes Leonhard, Moritz Ladenburger, Frederik Alexander, Vishisht Choudhary, Fabio Laredo, Giacomo Masserdotti, Thorben Prein, Carsten Marr et al. Published: 2026-07-05 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2607.04353v1 Hierarchical structure is common in image data, where fine-grained clusters often merge into larger, coarser semantic groups. In biological cell images, current self-supervised learning models often suppress this hierarchy, as coarse factors such as imaging modality can obscure finer morphological attributes in the latent space. We propose a hierarchy-aware self-supervised training framework to address this problem. Our method combines two components: a distillation framework with a segmentation teacher to improve morphological awareness in the latent space, and a hierarchy-aware contrastive loss based on HDBSCAN to improve decision boundaries between closely related subtypes at different hierarchical levels. Together, these components reduce the tendency of self-supervised learning to overemphasize coarse factors and instead align embeddings with semantic and morphological cues. This yields biologically meaningful sub-clusters driven by fine morphological detail. We train and evaluate our method on a curated corpus of 2.3 million single cells aggregated from 20 microscopy datasets, both labeled and unlabeled, covering 208 cell classes. Our method improves over baseline and counterpart methods, increasing average top-K accuracy by 2.8%, top-9 retrieval on the dataset with the deepest hierarchy by 6.3%, and downstream F1-score for biologically relevant drug classification from perturbed cell morphology by 7.8%. --- ### Last-Meter Precision Navigation for UAVs: A Diffusion-Refined Aerial Visual Servoing Approach Authors: Yaxuan Li, Jiarui Zeng, Shaofei Huang, Zhedong Zheng Published: 2026-07-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.04352v1 In this work, we study the last-meter precision navigation for UAVs, e.g., autonomously reaching a target within the final 10 meters using monocular vision. This task is challenging due to scale ambiguity, rotation discontinuities, and the need for fine-grained spatial reasoning. Existing methods often fail under large viewpoint changes or lack generalization to unseen environments. To this end, we propose DreamNav, a coarse-to-fine diffusion-refined aerial visual servoing framework. In the first coarse-estimation stage, a robust regression policy employs a trigonometric parameterization to predict rotation by jointly modeling sine and cosine components, effectively mitigating optimization instabilities caused by angular periodicity. Given this coarse estimate, the second diffusion-refined stage utilizes a pre-trained world model to simulate future visual observations for candidate actions, selecting the trajectory that minimizes visual discrepancy with the target through a process of visual imagination. To support rigorous evaluation, we contribute PairUAV, a large-scale benchmark comprising 4.8 million image pairs across 72 scenes, curated from the University-1652 dataset. Extensive experiments show DreamNav outperforms strong visual servoing and foundation model baselines in accuracy and generalization, with zero-shot transfer to unseen scenes. --- ### WPG-MoE: Weak-Prior-Guided Dense Mixture-of-Experts for User-Level Social Media Depression Detection Authors: Xian Li, Yuanhe Tian, Yang Yang, Guoqing Wang, Yan Song Published: 2026-07-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2607.04350v1 Online social media posts provide scalable signals for early depression screening, and recent studies mainly improve pre-classification evidence through risk-post selection, symptom grounding, and clinically informed feature construction. However, these screening-stage designs often leave final decisions to a single detector, overlooking how users heterogeneously express depressive risk after screening. A monolithic classifier must average across heterogeneous users, which may dilute localized evidence and cause misclassification, especially for non-self-disclosing users. To address this issue, we propose WPG-MoE, a weak-prior-guided dense mixture-of-experts framework built on a shared large language model (LLM) backbone. WPG-MoE derives user-level weak semantic priors to softly route users to experts matched to different evidence layouts. We formulate this process as learning using privileged information (LUPI): rich LLM-extracted structured evidence guides training-time routing, while inference retains only Patient Health Questionnaire-9 (PHQ-9) template screening and the deployable backbone. Experiments on Chinese and English datasets show that WPG-MoE outperforms strong baselines with interpretable routing behavior. --- ### IRIS: An Intelligent Vision-Language System for Ocular Surface Diseases via Topic Tree and Scene-Driven VQA Generation Authors: Hao Wei, Wenjin Qi, Dasen Dai, Minqing Zhang, Wu Yuan Published: 2026-07-05 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2607.04344v1 While Large Vision-Language Models (VLMs) demonstrate remarkable generic capabilities, their clinical reasoning in specialized domains like ocular surface diseases (OSDs) is severely hindered by a paucity of high-fidelity, multimodal instruction-tuning data. To dismantle this data bottleneck, we introduce IRIS, an Intelligent Recognition and Interaction System tailored for fine-grained OSD understanding via external eye photography. First, we curate IRIS-120K, the largest and most comprehensive OSD visual question-answering (VQA) dataset to date. Crucially, to overcome the semantic shallowness of conventional image-caption pairs, we propose a synergistic data generation paradigm to explicitly inject clinical priors. Our data engine operates via a dual-branch framework: 1) a Topic Finding Tree (TFT) that hierarchically anchors visual features to precise anatomical and pathological concepts, enforcing rigorous medical deduction logic; and 2) a Scene-driven strategy that synthesizes role-adaptive clinical dialogues to ensure pragmatic generalization. By explicitly aligning a compact 4B-parameter VLM on this structurally enriched corpus, IRIS achieves state-of-the-art performance, comprehensively outperforming both generalist and specialized medical VLMs with up to 34B parameters. Our findings underscore that structured knowledge injection profoundly prevails over sheer parameter scaling, unlocking the potential for resource-efficient, expert-level AI deployment on mobile edge devices for scalable OSD screening. Code, datasets, and model weights will be publicly released by this repo. --- ### How to Build Digital Humans? From Priors to Photorealistic Avatars Authors: Wojciech Zielonka, Tobias Kirschstein, Timo Bolkart, Simon Giebenhain, Vanessa Sklyarova, Xiang Deng, Donglai Xiang, Shunsuke Saito, Yebin Liu, Matthias Niessner, Justus Thies Published: 2026-07-05 Categories: cs.GR, cs.CL, cs.CV, eess.IV arXiv: https://arxiv.org/abs/2607.04341v1 This state-of-the-art report provides an overview of controllable 3D human avatar creation. We describe current 3D avatar systems, which typically consist of three stages: (i) learning priors of human appearance and motion, (ii) creating a personalized avatar, and (iii) animating the avatar. To limit the scope, we focus on the prior learning and avatar creation stages. We define current avatar representations and introduce a taxonomy that categorizes existing work along multiple axes, including body regions and employed priors. We review methods for full-body and head avatars, as well as layered representations that decompose the body into components such as hands, hair, and garments. Finally, we outline common underlying principles, reference key literature for newcomers, and discuss open challenges and future research directions. --- ### One Framework for All: Cross-Modal Membership Inference for Generative Models Authors: Dayong Ye, Tainqing Zhu, Kun Gao, Junhao Liu, Yichuan Chen, Shuai Zhou, Hengzhu Liu, Bo Liu, Wanlei Zhou Published: 2026-07-05 Categories: cs.LG, cs.AI, cs.CR arXiv: https://arxiv.org/abs/2607.04339v1 Large generative models across text-to-text, text-to-image, and image-to-text modalities have been shown to pose significant privacy risks. One fundamental threat is membership inference attacks (MIA), which aim to determine whether a given data point was used in a model's training set. Although prior work has investigated MIAs against these three classes of generative models, existing approaches treat them in isolation and are not cross-applicable, thereby limiting their real-world utility. To address this limitation, we present the first comprehensive study of a unified membership inference framework that applies across text-to-text, text-to-image, and image-to-text modalities. Our approach is grounded in a key modality-agnostic observation: the output distribution of a generative model can approximate its training data distribution. Leveraging this property, we model the distributions of model-generated outputs and auxiliary non-member samples in a shared embedding space, and perform membership inference via likelihood ratio testing. We conduct extensive experiments in a strict black-box setting under both partial-knowledge and zero-knowledge threat models, and evaluate membership inference against both fine-tuning and pre-training data. Experimental results demonstrate our approach's superior performance in comparison to existing state-of-the-art methods, which are typically optimized for a single model class. --- ### Server-side Anti-cheat in FPS games for Aimbot detection using Deep learning and Machine learning Authors: Siddhesh A. Dhinge, Shubham G. Sukum, Harsh S. Ranjane, Ruturajsingh R. Rajput, Jyoti H. Jadhav Published: 2026-07-05 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2607.04336v1 Modern video games are becoming more complex day by day. Most of these modern games are multiplayer first-person shooter (FPS) games. The rising popularity of FPS games emphasizes the need to combat cheating for fair and enjoyable gaming. As the number of players using cheating techniques like aimbots, wallhacks, and speed hacks is also increasing, we need a way to detect players who are using cheating tools to gain an unfair advantage over regular players. In this system, we focus exclusively on detecting aimbot cheats. Players who use aimbot cheats generally do not prioritize other aspects of the game. To distinguish between regular and cheating players, we identify specific features encompassing time-series data such as aim velocity, number of shots, distance to target, and more, along with behavioral data such as utility usage, player movement, and other gameplay patterns. Utilizing these features, we construct a server-side aimbot detection classifier named 'YAACS'. YAACS comprises a parser, a deep learning model, and intermediary connection utilities designed for integration with the game server. The proposed system achieves a classification accuracy of 88.6% with a false positive rate of 0.97% using a Stacked LSTM with Dense layers trained on sequences of 128 ticks (Tick Delta Negative=56, Tick Delta Positive=24), outperforming the Decision Tree baseline which achieves a higher accuracy of 96.2% but at a false positive rate of 2.68%, 2.76x worse than the best LSTM configuration. These results demonstrate that incorporating temporal context through sequence modelling is critical for minimising false accusations in FPS cheat detection. --- ### Do GUI Agents Believe Their Eyes? Diagnosing State-Belief Reliance on Pixels versus Structure Authors: Guijia Zhang, Yuxun Chen, Yuheng Qi, Harry Yang Published: 2026-07-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.04334v2 Multimodal GUI agents read an interface through two redundant channels: the rendered pixels of a screenshot and a serialized structure such as a document object model or accessibility tree. Before acting, an agent forms a belief about the current interface state, but existing benchmarks score task success, element grounding, or attack resistance and do not ask whether that belief is drawn from the pixels. We formalize visual state reliance, the attribution of a state belief to pixels, structure, or priors, and measure it with paired single-channel interventions over 735 probes spanning real web, mobile, and desktop interfaces, of which 225 are zero-edit divergences mined from live production websites, all scored by deterministic forced choice with no model judge. Our central metric is the Perception-Fusion Gap (PFG), the fraction of probes a model perceives correctly yet resolves toward structure under conflict; a stricter variant that re-verifies perception on a tight crop of the target region leaves the gap intact. Across models from four vendors, textual state beliefs defer to structure while image-only accuracy stays near ceiling, and on unedited stale snapshots from live pages the same models follow the outdated structure on up to 0.88 of probes. A white-box ablation traces the textual effect to a single copied structural value, and gradient attribution shows the visual evidence is processed yet overridden. In live multi-step environments, one mis-sourced belief at the first step compounds into task failure with a self-recovery rate of at most 0.03. Comparing four mitigations on identical probes, prompt-level cues fail at the action level, certificate checks buy safety with refusals, and a training-free consistency gate is alone in reducing both hijack and task error. Visual state reliance thus gives a measurable diagnostic of whether agent state beliefs are visually grounded. --- ### Structure-Specific Representational Priors Causally Control the Grokking Delay Authors: Gunner Levi Howe Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04333v3 Grokking -- generalization long after training-set interpolation -- has been accelerated by structure-agnostic interventions (gradient filtering, weight-norm clamping, geometric penalties). Whether the delay specifically measures the time to form task-structured representations has remained observational. We test it causally by injecting representational priors of varying content into a one-layer transformer learning modular addition, via a supervised-contrastive loss whose positives encode (i) the task's true structure ($(a+b) \bmod p$), (ii) a coherent-but-wrong sibling ($(a-b) \bmod p$), or (iii) a random partition -- all with identical loss form, strength, class sizes, and geometry. Whether generalization occurs follows a clean gradation: true 22/30 runs, sibling (same periodic features, wrong combination) 14/15, random (only memorizable) 0/20 (Fisher $p=1.3\times10^{-7}$). A weight-norm-matched control replaying the norm trajectory onto plain cross-entropy generalizes 0/15, ruling out the norm as mediator. Probes show structure formation precedes and predicts generalization in all runs. Only the true structure also accelerates grokking (up to $2.75\times$), but this is dose-dependent and bimodal. We then confirm the mechanism by prediction: because the acceleration is gated by a weight-norm side-effect, clamping the norm during training yields a reliable, standalone accelerator with a median $8.6\times$ speedup (up to $22\times$ on the fastest seeds, under 1000 epochs), growing monotonically as the norm is held lower; the residual stalls also vanish, though significant only pooled over the two mitigations run at both strengths ($0/40$ vs $6/20$, $p=7.7\times10^{-4}$), not per method. The grokking delay is, causally, the time to form the right representational structure -- decided at the level of features, not labels. --- ### On the effectiveness of reward functions in reinforcement learning for confidence calibration of large language models Authors: Chee Heng Tan, Zhuoyi Lin, Mehul Motani, Wee Sun Lee Published: 2026-07-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2607.04332v1 In this paper, we consider the setting where large language models (LLMs) are trained using reinforcement learning (RL) to simultaneously improve reasoning accuracy and verbalize its confidence. Our reward scheme uses two functions for rewarding confidence verbalized by the LLM: one when the LLM is correct and a different one when the LLM is incorrect. With a poorly designed reward scheme, the LLM may be incentivized to answer incorrectly so that it can be confident that its answer is indeed incorrect, a phenomenon that we call confidence reward hacking. We propose the concept of non-hackable confidence reward schemes and define a spectrum of such reward schemes for RL confidence calibration training in LLMs. We demonstrate that selective confidence reward hacking can occur in practical datasets with reward schemes that are not designed to be non-hackable. We also demonstrate that the reward scheme with the best calibration to accuracy tradeoff depends on the dataset and the application, and propose using the reward scheme as a hyperparameter to optimize the tradeoffs in accordance to what is important for the application. The code of our experiments is available in https://anonymous.4open.science/r/rl-confidence-calibration-9ED4/README.md. --- ### Agent-driven Long-tail Simulation for Autonomous Driving Authors: Junru Gu, Lijin Yang, Jianing Huang, Shu Liu, Zhongzhan Huang, Hang Zhao Published: 2026-07-05 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2607.04331v1 Evaluating autonomous driving systems in closed-loop settings requires realistic and interactive simulation, yet existing simulators largely rely on log replay or rule-based agents, limiting behavioral diversity and long-tail coverage. We propose an agent-driven simulation framework in which surrounding road participants are controlled by instruction-following large language models through a structured action interface, enabling intentional and reactive behaviors while preserving physical plausibility. Furthermore, we introduce SemanticPlan, a benchmark of closed-loop planning in long-tail and semantically rich scenarios that augment real nuPlan scenes with multiple interactive agents following diverse language instructions. Evaluation results show that state-of-the-art planners still struggle to consistently achieve safe and effective task completion, suggesting that these long-tail scenarios remain challenging. --- ### Framework and Multi-modal Dataset for Roadwork Zone Detection and Geo-localization Authors: Zhiran Yan, Yutong Xin, S Shyam Shenoi, Rui Song, Gordon Elger Published: 2026-07-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.04330v1 Autonomous vehicles often rely on high-definition (HD) maps for navigation; however, these maps are not frequently updated and often lack semi-static information, such as temporary roadwork zones, which can significantly alter the road network. This limitation underscores the urgent need for an accurate global position of roadwork zones. However, the absence of publicly available datasets for evaluating roadwork zone detection and geo-localization models has hindered the development of reliable autonomous driving systems. To address this challenge, we propose the Roadwork Zone Detection and Geo-localization (RZDG) dataset, which includes both simulated and real-world data, providing multimodal sensor inputs along with comprehensive annotations. The dataset supports multiple perception tasks, including image semantic segmentation, 3D object detection, and object geo-localization. In addition, we introduce a tracker-based roadwork zone detection and geo-localization (RZDG) pipeline, an extension of AB3DMOT, for accurate object geo-localization in roadwork zones. We benchmark our approach on the RZDG dataset, demonstrating its effectiveness in detecting roadwork zones and transforming object positions from the local coordinate system to the global coordinate system. A prediction is considered a true positive (TP) if its estimated position falls within one meter of the ground truth. Our experimental results show that our approach achieves high accuracy on both real and simulated data. Specifically, we report: Precision: 0.565 (real) / 0.615 (simulated) Recall: 0.898 (real) / 0.809 (simulated) F1-score: 0.597 (real) / 0.665 (simulated). --- ### HAS-Bench: Evaluating LLM-Based Human-Agent Systems under Configurable Human Participation Authors: Yaozu Wu, Wei-Chieh Huang, Jizhou Guo, Dongyuan Li, Renhe Jiang, Henry Peng Zou, Chunyu Miao, Shanghao Li, Weizhi Zhang, WeiWei Ye, Yankai Chen, Meng Zhang et al. Published: 2026-07-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2607.04329v1 Large language models increasingly operate in settings where humans are active collaborators rather than passive task providers. We introduce HAS-Framework, a graph-based framework that represents humans and LLM-powered agents as first-class participants with explicit roles, permissions, communication paths, and action authority. Building on this framework, HAS-Bench evaluates Human-Agent Systems under configurable human participation across agency levels, interaction channels, and persona policies. The benchmark measures both task outcomes and process-level collaboration behavior, including clarification quality, feedback utilization, control calibration, safety, initiative, and interaction cost. Experiments across six domains show that human participation can substantially improve task completion and failure recovery, but the gains depend on when, how, and by whom human input is exercised. --- ### Legible-by-Construction: Attention and End-to-End Transformers Authors: Mark Oskin Published: 2026-07-05 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2607.04319v1 A companion paper showed that a transformer's feed-forward layer can be rebuilt from explicit fuzzy set operations - intersection, set-difference, and a self-forgetting sequence quantifier - so its hidden units read as named logical operators at no cost to language-model quality. That left the other half of the transformer opaque. Here we carry the same idea into attention and join the two into one model. The mechanism is minimal: a head's value is passed through a sigmoid, so each value channel becomes a readable detector of whether a feature holds at a token. This adds no parameters and leaves the standard head otherwise untouched. A Boolean variant goes further, restructuring the value into an explicit within-token intersection and negation-capable set-difference. In both designs the output projection is left free, not tied to the vocabulary, which is the load-bearing decision: bounding what a head detects while leaving what it writes unconstrained yields selective detectors, whereas constraining the write does not. A bounded value is shaped into a readable detector by two selectivity pressures - one for sparse firing, one for decisive firing at the rails - and which a design wants is not universal. Across five specialized-attention designs at 125M parameters, 44 to 62 percent of value channels become crisp, contextually selective detectors, and their legibility rises with depth rather than crystallizing only on punctuation. Language-model quality is at parity with a conventional baseline. Finally, we couple the Boolean attention to the legible feed-forward layer and train an end-to-end legible-by-construction language model at benchmark parity: its feed-forward units are named set and quantifier operations throughout, and we can take a token it generates and read the named units that compose to produce it. --- ### Fixed-Confidence Best-Arm Identification for Causal Mediation Analysis Authors: Harsh Shrivastava, Yuta Kawakami, Junpei Komiyama, Jin Tian Published: 2026-07-05 Categories: stat.ML, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2607.04315v1 This paper studies the problem of identifying the treatment that maximizes the expected natural direct potential outcome (NDPO), which captures the potential outcome of an intervention while excluding the pathway transmitted through a mediator that researchers may wish to remove from evaluation. We first establish population-level identification of the expected NDPO in a causal bandit setting using observable interventional distributions. We then develop a fixed-confidence best-arm identification (BAI) algorithm based on the Track-and-Stop (TaS) framework, employing a cutting-set method to solve the resulting semi-infinite optimization problem. The proposed algorithm achieves sample-efficient identification with a high-probability correctness guarantee. We prove that it satisfies $δ$-correctness and asymptotic optimality. Finally, we validate the approach through empirical evaluations on a large-scale real-world advertising dataset (IPinYou). --- ### Aura: Consistent Multi-Subject Video Generation via VLM-Grounded Semantic Alignment Authors: Zixiang Zhou, Zhentao Yu, Yifeng Ma, Hongmei Wang, Wenqing Yu, Cong Wang, Zilin Yang, Rui Chen, Jiarong Ou, Yezhou Liu, Yuan Zhou, Qinglin Lu Published: 2026-07-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.04311v2 Subject-driven and multi-element video generation are central to controllable video synthesis, but existing methods still struggle to preserve identity consistency and model complex relationships among multiple subjects. In this paper, we propose Aura, a unified framework for high-fidelity and identity-consistent video generation. To better capture scene dynamics and subject interactions, we introduce AI director-level captions that provide dense and structured descriptions of video content. We further leverage a vision-language model (VLM) with learnable queries to extract multimodal semantic features from textual and visual references, covering both global semantics and fine-grained visual cues. To bridge the representational gap between the VLM and the Diffusion Transformer (DiT), we design a two-stage alignment strategy that progressively maps VLM features into the DiT feature space. For visual conditioning, we adopt token concatenation to inject reference information directly into the generation process. To distinguish heterogeneous subject types and reduce common copy-paste artifacts, we develop a subject-aware RoPE-Shift mechanism. To further differentiate reference images of different categories, we introduce subject-aware learnable tokens. In addition, we introduce Memory Tokens to balance the training signal across examples with different numbers of reference subjects. During inference, Progressive-APG (Adaptive Prompt Guidance) further alleviates oversaturation and improves semantic alignment with user prompts. Finally, we build a high-quality video-subject image dataset through a dedicated data construction pipeline. Extensive experiments show that our method achieves state-of-the-art performance on both single-subject generation and more challenging multi-element scenarios. --- ### SAD-LoRA: Spectral Alignment for Low-Rank Knowledge Distillation Authors: Omer Tariq, Syed Muhammad Raza, Jeongbae Son Published: 2026-07-05 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2607.04306v1 Distilling a fine-tuned teacher into a LoRA-adapted student is a standard recipe for parameter-efficient compression, but output-level KD does not explicitly control which rank-$r$ weight subspace the adapter occupies. We propose \textbf{SAD-LoRA} (\textbf{S}pectral \textbf{A}lignment \textbf{D}istillation), which selects this subspace from the data-weighted student-space reference update $\DWT\Sigx^{1/2}$ and maintains it during training via a differentiable principal-angle loss on $\colspan(B)$. We show that the data-weighted distillation error decomposes exactly into subspace misalignment, within-subspace coefficient mismatch, and irreducible rank residual; standard KD can affect the first term only indirectly through output gradients. On controlled synthetic problems with a flat teacher spectrum, SAD-LoRA reduces the subspace-misalignment term from $51\%$ to nearly zero and lifts final subspace alignment from $0.49$ to $1.00$. On RoBERTa-large to RoBERTa-base distillation across six GLUE tasks, SAD-LoRA improves rank efficiency: at $r{=}4$, it matches or beats the strongest included spectral baseline on five of six tasks, and at $r{=}8$ it gives the best result on SST-2 and CoLA. Ablations identify subspace alignment as the load-bearing component, while coefficient matching is auxiliary. --- ### Road-Aware Anomaly Segmentation with Query-Guided Polygons and CLIP in Autonomous Driving Authors: Zhiran Yan, Gordon Elger Published: 2026-07-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.04304v1 Traditional semantic segmentation models operate under a closed-set assumption and struggle to recognize unknown or unexpected objects-an essential capability for autonomous driving. As a result, such models often misclassify or overlook out-of-distribution (OOD) road anomalies, posing safety risks in open-world environments. We present a lightweight, postprocessing, road-aware anomaly segmentation framework that requires no retraining, no OOD data, and no auxiliary supervision. Our approach builds on a mask transformer-based segmentation network by exploiting query-level mask confidence and deriving a polygonal road prior to detect gap regions that may correspond to anomalies. To further suppress false positives, we introduce a CLIP-based zero-shot semantic filtering module using in-distribution prompts, with optional generalized OOD prompts. By jointly leveraging spatial priors and semantic verification, our framework produces robust and interpretable anomaly predictions. Evaluation on three public benchmarks-Fishyscapes, SMIYC, and RoadAnomaly-shows consistently strong performance. In particular, our method outperforms the training-free baseline Maskomaly on most metrics and achieves the highest AP on Fishyscapes LostAndFound. These results demonstrate the practicality and deployability of our approach for real-world autonomous driving systems. --- ### AquaStereo: Enabling Underwater Stereo Matching via Depth-Conditioned Diffusion and Geometry Self-Distillation Authors: Qizhe Wei, Yingping Liang, Shaodi You, Ying Fu Published: 2026-07-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2607.04303v1 Learning-based stereo matching models struggle in underwater environments due to scarce in-domain data and the difficulty of extracting discriminative correspondences from degraded imagery. In this work, we present $\textbf{AquaStereo}$, a perception-enhanced framework with a data simulation pipeline and a self-distillation strategy that jointly address data scarcity and feature degradation in underwater stereo matching. First, a depth-conditioned diffusion pipeline renders underwater stereo pairs while preserving binocular geometry, with a lightweight left-right consistency module ensuring geometric alignment. Training on this synthetic corpus effectively narrows the terrestrial-underwater gap and improves zero-shot robustness. Second, a frozen binocular teacher trained on clean terrestrial pairs guides a student exposed to rendered underwater pairs with perturbations. A stage-weighted sequence loss is performed to align the student's disparities with the teacher's geometry, while a clean-branch supervision with shared pseudo targets prevents scale drift. To further enhance feature stability under turbidity and low texture, we introduce learnable perception frames, a perception-enhanced feature formulation that constructs robust matching descriptors by fusing temporal cues from two auxiliary views encoded by a video backbone with semantic features extracted by a strong image encoder. Extensive experiments demonstrate that $\textbf{AquaStereo}$ substantially improves robustness and zero-shot generalization in challenging underwater scenarios. The code is available at https://github.com/qz-wei/AquaStereo. --- ### HiFA4: Training-Free 4-bit FlashAttention on Ascend HIF4 NPUs for LLM Inference Authors: Hui Dong, Yanzhao Li, Jie Gao, Chunlu Li, Zhiyuan Zhang, Yupeng Sun, Zhenyuan Chen, Zhiqiang Zou Published: 2026-07-05 Categories: cs.LG, cs.AI, cs.AR, cs.CL, cs.PF arXiv: https://arxiv.org/abs/2607.04302v1 We present HiFA4, a post-training operator-level design that executes both QK^T and PV in FlashAttention as 4-bit HIF4 Cube GEMMs for LLM inference on Ascend NPUs, while maintaining the online softmax state in FP16. To our knowledge, HiFA4 is the first Ascend-HIF4-targeted design of this kind evaluated on standard NLP benchmarks. HiFA4 combines two mechanisms. Smooth-QK applies a calibration-static per-channel equivalent rescaling to Q and K after RoPE, transferring quantization difficulty from K to Q without per-tile online reduction at inference. P-Reordering accumulates the softmax normalizer from the same quantized attention weights P_hat used in the PV GEMM, rather than from a higher-precision reconstruction. We show that this inconsistent formulation introduces a coherent output-scaling error, and validate the effect on a Qwen3-8B Layer-0 MMLU trace, where all 3.6M measured attention tiles exhibit net probability-mass loss with median epsilon_bar = -0.064. P-Reordering also allows the normalizer to be fused into the PV Cube GEMM. Across five LLMs, HiFA4 reduces quantization-induced decision drift. On Qwen3-8B, it recovers 37.5% of the accuracy gap introduced by direct HIF4 quantization, narrows the sample-weighted accuracy loss from 1.12 pp to 0.70 pp, reduces BF16-inconsistent MMLU predictions from 16.3% to 8.2%, and cuts MMLU accuracy regressions by 57% (1071 to 465). On Gemma2-9B, mild smoothing keeps HiFA4 within 0.7 pp of BF16 while reducing MMLU regressions by 27%. On LLaMA3.1-8B, Mistral-7B, and Phi-4B, where Smooth-QK is disabled, P-Reordering with the adopted Q-Mean auxiliary still reduces full-set MMLU regressions by 41-52%. A preliminary instruction-scheduling analysis projects a 35.4% critical-path latency reduction relative to BF16 by fusing the softmax normalizer into the PV Cube GEMM; on-hardware validation is left to future work. --- ### CausalGame: Benchmarking Causal Thinking of LLM Agents in Games Authors: Zhenhao Chen, Yongqiang Chen, Chenxi Liu, Junchi Yu, Xiangchen Song, Zijian Li, Jialin Li, Philip Torr, Bo Han, Kun Zhang Published: 2026-07-05 Categories: cs.CL, cs.AI, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2607.04293v1 Building AI Scientist agents with Large Language Models (LLMs) has recently attracted growing attention. Since scientific discovery fundamentally relies on uncovering causal relationships from observations, the capability of causal thinking, i.e., distinguishing causation from correlation and recognizing hidden biases, is essential to LLM agents. Although a number of benchmarks exist for AI Scientists, none explicitly incorporate challenges from selection bias, measurement error, and hidden confounders that widely exist in real-world scientific discovery. To this end, we present CausalGame, a benchmark that evaluates the causal thinking capabilities of LLM agents through interactive games. CausalGame asks LLM agents to actively design experimental protocols, collect observation data, and derive a final solution with an explanation report. To emulate realistic scientific discovery challenges, we design 14 scenarios that incorporate selection bias, measurement error, and hidden confounders. Across 30 LLM agents, none demonstrates reliable causal thinking: the best model reaches only 68.0% survival against analytical optima of 78-85%, and merely 5-7% of sessions receive credits on the causal-reasoning rubrics. CausalGame provides a scalable and controlled testbed for evaluating the causal thinking of AI Scientist agents. ---