# Research Papers — 2026-08 · bundle page 204 of 240 > Full-text markdown of papers #10151–10200 of 11995 published in 2026-08, newest first. AI-readable mirrors of the full 11995-paper 2026-08 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-08/page/204 [← Previous bundle](https://aipod.dev/papers/archive/2026-08/page/203.md) [Next bundle →](https://aipod.dev/papers/archive/2026-08/page/205.md) ### ReGround: Restoring Visual Grounding in Multi-Step Reasoning through Self-Diagnosis and Visual Re-Examination Authors: Lei Peng, Shuai Lv, Wei Hu Published: 2026-08-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.04385v1 Vision-Language Models (VLMs) often lose visual grounding during multi-step reasoning: as reasoning chains grow longer, later inference steps rely increasingly on language priors rather than image evidence. We identify a consistent benchmark-level signature associated with this degradation: across 2,510 re-examined samples from four benchmarks, attention entropy over image tokens typically decreases during Round 1 and rises again after image re-injection. However, we find that effective visual re-examination requires two complementary ingredients: image re-injection and targeted self-diagnosis. Without targeted diagnosis, re-examination can even hurt performance, whereas accurate self-diagnosis yields substantial gains -- a swing of several points on key benchmarks, indicating that diagnostic quality is a key factor in whether re-examination helps or hurts in our setting. We present ReGround, a two-stage framework that teaches VLMs to self-diagnose grounding failures and selectively re-examine visual evidence, without architectural modifications or external tools. Through capability bootstrapping, a stronger variant from the same model family provides diagnostic scaffolding only during data construction, while the policy model learns to diagnose autonomously at inference time and retains most of the assisted gains. Experiments on eight benchmarks across two VLM backbones demonstrate consistent gains, especially on visually intensive multi-step reasoning tasks, while incurring only modest inference overhead relative to tool-augmented baselines. Project page: https://sespoir.github.io/reground-page/ . Code: https://github.com/sespoir/ReGround . --- ### Improving Auto-Design of Neural PDE Solvers with a Domain-Specific Language Authors: Shengxin Kong, Liwen Xu, Jingwen Fu Published: 2026-08-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.04384v2 Neural PDE solver auto-design is fundamentally a search-space representation problem. In the space of unrestricted Python programs, valid solvers form an extremely sparse subset: most candidate programs are syntactically incorrect, semantically incompatible, or numerically unstable. Direct code generation therefore forces an LLM to spend most of its search capacity navigating implementation failures rather than reasoning about solver quality. ADSL-PDE addresses this challenge by introducing a structured search state between solver concepts and executable code. It represents the functional decisions that determine a neural PDE solver (architecture, physical constraints, objectives, sampling, and optimization) while abstracting away low-level implementation details. A deterministic compiler maps each valid search state to an executable solver. In effect, ADSL-PDE reshapes the search space: it removes large regions of invalid programs, increases the density of meaningful candidates, and preserves the compositional freedom needed to discover previously unseen designs. Solver evolution can thus operate over design decisions rather than code artifacts. Built on this representation, our evolutionary agent iteratively proposes, evaluates, and refines solver search states using empirical feedback. Across multiple PDE benchmarks, ADSL-PDE improves both search efficiency and optimization stability, achieving an improvement of more than 52% within the first ten evolution iterations. These results suggest a broader principle for LLM-driven auto-design: effective agents do not merely require stronger reasoning, but rather a search representation that concentrates exploration on valid and consequential decisions. --- ### Improving Auto-Design of Neural PDE Solvers with a Domain-Specific Language Authors: Shengxin Kong, Liwen Xu, Jingwen Fu Published: 2026-08-05 Categories: cs.AI arXiv: https://arxiv.org/abs/2608.04384v1 Neural PDE solver auto-design is fundamentally a search-space representation problem. In the space of unrestricted Python programs, valid solvers form an extremely sparse subset: most candidate programs are syntactically incorrect, semantically incompatible, or numerically unstable. Direct code generation therefore forces an LLM to spend most of its search capacity navigating implementation failures rather than reasoning about solver quality. ADSL-PDE addresses this challenge by introducing a structured search state between solver concepts and executable code. It represents the functional decisions that determine a neural PDE solver (architecture, physical constraints, objectives, sampling, and optimization) while abstracting away low-level implementation details. A deterministic compiler maps each valid search state to an executable solver. In effect, ADSL-PDE reshapes the search space: it removes large regions of invalid programs, increases the density of meaningful candidates, and preserves the compositional freedom needed to discover previously unseen designs. Solver evolution can thus operate over design decisions rather than code artifacts. Built on this representation, our evolutionary agent iteratively proposes, evaluates, and refines solver search states using empirical feedback. Across multiple PDE benchmarks, ADSL-PDE improves both search efficiency and optimization stability, achieving an improvement of more than 52% within the first ten evolution iterations. These results suggest a broader principle for LLM-driven auto-design: effective agents do not merely require stronger reasoning, but rather a search representation that concentrates exploration on valid and consequential decisions. --- ### Non-asymptotic implicit bias of logistic regression at early-stage gradient descent dynamics Authors: Han Bao Published: 2026-08-05 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.04382v1 Gradient descent has been of particular interest in modern machine learning beyond sole focus on optimization. Implicit bias emerging from optimization, though not being encoded by the learning objective, often prevents from overfitting to spurious patterns. A typical instance is the max-margin implicit bias of a linear classifier, widely established for exponentially tailed loss functions. Even after having a given dataset separated, the parameter vector continues to evolve towards the max-margin direction asymptotically along the gradient descent dynamics. This phenomenon corroborates a frequent empirical observation of "train longer, generalize better." However, the max-margin convergence is an asymptotic phenomenon, and what is worse, this asymptotic convergence rate is significantly slower than pure convex optimization. Even so, the parameter vector along gradient descent dynamics commonly correlates with the max-margin direction positively (though not exactly) within considerably fewer iterations than the asymptotic rate. By shedding another light on this classical problem, this work aims to understand the mechanism of this early-stage alignment phenomenon. Our theoretical results demonstrate that the parameter vector weakly aligns with the max-margin direction within $O(\exp(\exp(-δ)))$ iterations, where $δ>0$ is the permissible alignment error, which is shown to be tight. By tracking the radial and tangential flows, our proof operates on the alignment dynamics directly with dataset geometry and gets rid of the asymptotic expansion, which is a key insight to establishing faster weak alignment. --- ### NodeJEPA: Structure-Conditioned Latent Prediction for Node-Level Graph Self-Supervised Learning Authors: Tinghe Zhang, Jian Xu, Jiaheng Chen, Jiaxing Li, Yucheng Xiao, Qiang Wang Published: 2026-08-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.04381v1 Self-supervised learning on graphs is largely shaped by contrastive methods that depend on carefully designed augmentations, and by generative methods that reconstruct node attributes in the input space. Both paradigms can entangle representations with low-level input statistics rather than with relational structure. Joint-embedding predictive architectures (JEPA) instead learn by predicting latent targets rather than reconstructing inputs. Recent work has explored this idea for graph-level representation learning, but how to design JEPA-style objectives for node-level tasks, and which structural signals the predictor should condition on, remains less clear. We present NodeJEPA, a joint-embedding predictive architecture for node-level graph self-supervised learning. NodeJEPA masks structure-aware k-hop ego-subgraphs and trains a context encoder to predict the latent representations of the masked nodes. These targets come from an EMA-updated target encoder with stop-gradient. A structure-conditioned predictor integrates spectral and centrality descriptors through cross-attention. Variance, covariance, and Laplacian spectral regularizers help stabilize the embedding geometry, and an optional curriculum gradually increases masking difficulty during training. Because prediction occurs in latent space, NodeJEPA does not rely on input reconstruction or hand-crafted graph augmentations. We evaluate NodeJEPA on standard node classification benchmarks under linear probing and fine-tuning protocols, and conduct ablations on masking, prediction, and regularization design choices. Our study offers a practical recipe for node-level JEPA-style latent prediction on graphs, and clarifies when structural conditioning helps representation learning. Code, configurations, and evaluation scripts are publicly available at https://github.com/OliverZ-dot/Node-Jepa. --- ### Image Classification Using CNN-QNN Hybrid Model with Optimized Correlated Features Authors: Minseo Seong, Youngwook Kim Published: 2026-08-05 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2608.04379v1 We propose a method to optimize the correlation among convolutional neural network (CNN) features that are used as inputs to quantum neural network (QNN) to enhance image classification accuracy. Unlike prior approaches that employ orthogonal decomposition as preprocessing, we intentionally introduce correlated features that are more physically compatible with QNN. This design leverages the QNN's inherent ability to exploit quantum entanglement for representing correlated states-an advantage unavailable to classical neural networks. We hypothesize that aligning feature correlations with the entanglement structure of QNN improves binary classification performance. Based on a mathematical derivation of QNN outputs, Monte Carlo simulations indicate that an average correlation between features of 0.5 yields optimal classification accuracy. To validate this finding, we evaluate a quantum-classical hybrid model on three tasks: CIFAR-10 (automobile vs. truck), Fashion-MNIST (shirt vs. coat), and radar micro-Doppler signatures (robotic dogs vs. non-robots). To regulate feature correlations, we introduce a correlation-regularization term on the outputs of the CNN, driving the off-diagonal entries of the feature correlation matrix toward a target constant. Across all datasets, inducing intermediate correlation consistently improved accuracy compared to low, high, or unregulated correlations, while also reducing classification accuracy variance. These results demonstrate that imposing moderate feature correlations-without modifying the quantum circuit-enhances classification accuracy and stability by aligning feature statistics with the QNN's entanglement structure. This study highlights the potential of QNN to surpass the performance of classical classifiers as more qubits become available. --- ### Helping Music Co-Creation Agents 'Listen' Well: Hierarchical Self-Supervised World Models for Understanding and Generation Authors: Scott H. Hawley Published: 2026-08-05 Categories: cs.SD, cs.LG, eess.AS arXiv: https://arxiv.org/abs/2608.04378v1 Collaborative music agents need internal representations rich enough to support both understanding and generation, yet flexible enough for a workflow where the human retains agency. We present a hierarchical self-supervised ``world model'' for symbolic music: a 2.55M-parameter Swin V2 encoder trained on MIDI piano-roll images with JEPA-style objectives (pitch- and time-shift equivariance, masked embedding prediction, and a distributional regularizer), using no labels and no music-theory vocabulary. Probing the frozen embeddings shows that the level at which a musical property becomes decodable tracks its musical time scale: phrase boundaries are read off the coarsest levels, note density and harmonic detail off the finest. Temporal and phrase structure emerge from the self-supervised objectives alone, while harmonic content must be asked for; a small chord-supervision head raises joint chord recovery from .18 to .54, and key detection, which is never supervised, from .16 to .70. Following the Representation AutoEncoder paradigm, a conditional flow-matching model stands in for a trained decoder, flowing in pixel space from PCA-reduced conditioning: it reproduces a target window at pixel F1 $0.996$, and the same per-level conditioning dropout that controls how far variations stray also enables graphical prompting for masked inpainting with no inpainting-specific sampler. The pipeline runs on CPU producing a suggestion in $2.8$ s, or $0.6$ s on Apple MPS, which we demonstrate in a live interactive demo. In concert with an LLM-based brain, these capabilities supply the core of a collaborative music creation agent in service of, rather than in place of, human agency. --- ### Towards Trustworthy Hypergraph Neural Networks under Label Noise Authors: Mengyao Zhou, Zhiheng Zhou, Xiao Han, Guiying Yan Published: 2026-08-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.04377v1 Hypergraph neural networks (HGNNs) have demonstrated remarkable capabilities in processing complex higher-order relationships. However, their performance is highly dependent on labeled data, making them vulnerable to label noise. Despite advances in learning with label noise (LLN) and graph learning with label noise (GLN), noisy-label learning on hypergraphs remains underexplored. In this paper, we present a systematic study of hypergraph node classification under label noise. First, we adapt representative LLN and GLN methods to hypergraphs and evaluate them under a unified benchmark, revealing the limitations of existing robust learning strategies for hypergraphs. Building on this, we propose a new hypergraph robust framework, HyperTrust, which first estimates hyperedge trustworthiness through a pretraining-based, entropy-aware strategy, and then incorporates the HyperedgeBoost module to enhance reliable supervision by connecting unlabeled nodes to trustworthy hyperedges, as well as the HyperedgePrune module to suppress noisy propagation by removing untrustworthy node-hyperedge incidences. Finally, two modules work collaboratively to adjust the hypergraph structure and generate final predictions. Extensive experiments and theoretical analysis demonstrate the effectiveness and robustness of HyperTrust on multiple hypergraph datasets under various noisy settings. Our work provides a unified benchmark and an effective solution for hypergraph learning with label noise and lays a foundation for future research in this direction. --- ### FinReportBench: Measuring and Improving Institution-Grade Financial Report Generation Authors: Yinghao Tang, Tan Zhenwei, Yiyao Wang, Wanli Gu, Xiaolu Zhang, Jun Zhou, Wei Chen Published: 2026-08-05 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2608.04374v1 Large language models can produce fluent financial analysis, but fluency alone does not establish whether a report is suitable for institutional delivery. We introduce FinReportBench, an expert-grounded benchmark for measuring and improving institution-grade financial report generation. Expert review reveals recurring gaps in report identity, institutional components, source discipline, and visual delivery. We derive a 35-item rubric through expert partial orders, multimodal evidence, and audits of decision boundaries, covering deliverability, report identity, and institutional completeness. Starting from 10,000 balanced Chinese and English financial-research source records, we curate 244 bilingual tasks across three research objects and two input tiers. Each task separates the public query, reconstructed research trajectory, and hidden source packet. Three independent judge families reproduce the expert partial order at near-ceiling rates, showing that bounded, observable criteria support reliable evaluation. Across nine model families, basic deliverability is nearly saturated, while report identity and institutional completeness remain the primary bottlenecks. The largest cross-model gaps concern generation-trace control, information density, and data discipline rather than basic report framing. We then use benchmark-guided skill distillation to turn recurrent failures into reusable generation and self-review constraints. Across five model families, the evolved skill improves mean G1 by 33.85 points and mean G2 by 13.83 points over paired no-skill runs while preserving G0 for every pair. Code and benchmark artifacts are available at https://github.com/MisterBrookT/finreportbench. --- ### EvtGraph: Event-Adaptive Compression for Sparse Temporal Graph Learning in Multimodal Time Series Authors: Ziqian Wang, Tingxiong Xiao, Yuxiao Cheng, Jinli Suo Published: 2026-08-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.04368v1 Multimodal temporal data are inherently irregular and uneven in information density, yet most models rely on uniform discretization, leading to inefficient representations. We propose \textbf{EvtGraph}, a unified framework that aligns computation with temporal salience under explicit budget constraints. EvtGraph reparameterizes sequences into event-level tokens via event-adaptive compression (EAMC), selects a compact subset with a node budget (NBC), and performs temporally constrained sparse graph reasoning (T2SG). This transforms dense sequences into structured computation over salient events, reducing complexity while preserving critical transitions. We show that this design provides a practical mechanism for allocating representational capacity under a fixed budget, yielding a consistent performance--efficiency trade-off, where a small budget is often sufficient in practice. Experiments on multimodal clinical (MIMIC-IV + CXR) and cross-domain benchmarks demonstrate that EvtGraph outperforms both Transformer-based and recurrent baselines while significantly improving efficiency. These results suggest that budget-constrained event-centric representation provides a general paradigm for learning from high-redundancy temporal data. --- ### Combating Knowledge Corruption in Agent Systems: A Byzantine-Tolerant Secure Collaborative RAG Framework Authors: Zhaoqi Wang, Daqing He, Zijian Zhang, Ye Liu, Jiamou Liu, Zhirui Zeng, Zhan Qin, Zhen Li, Xin Li, Hongwei Yao, Jincheng An, Yong Liu et al. Published: 2026-08-05 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2608.04366v1 While retrieval-augmented generation systems partially address the hallucination issues in large language models, it also introduces new vulnerabilities to knowledge corruption attacks. Adversaries exploit these vulnerabilities by poisoning documents provided by RAG system to manipulate LLM outputs. To counter this threat, we propose SecureCollaRAG, a Byzantine-tolerant collaborative RAG framework leveraging Multi-source Knowledge Validation Mechanism. Our approach enables agent system to securely verify document provenance through dynamic GNN-based credibility scoring, effectively preventing stealthy knowledge corruption attacks while preserving essential domain knowledge integrity. Through extensive evaluations and formal analysis, we demonstrate that SecureCollaRAG maintains robustness against attackers under non-IID data distributions. --- ### Manipulation-Proof Oblivious Audits against Deceptive Model Providers Authors: Augustin Godinot, Sofiane Azogagh, Julien Ferry, Sébastien Gambs Published: 2026-08-05 Categories: cs.LG, cs.CR, cs.CY arXiv: https://arxiv.org/abs/2608.04365v1 Audits have emerged as a critical instrument for algorithmic governance, providing a mechanism for external scrutiny and governance of machine learning models. However, ensuring the integrity of such assessments remains a challenging issue. For instance in regulatory contexts, audits are typically declared or easily detected, thus enabling model providers to manipulate the process, whether intentionally or inadvertently. This vulnerability is particularly acute in the context of fairness evaluations, in which providers can often infer sensitive attributes and strategically equalize allocation rates between groups to satisfy fairness metrics. In this paper, we introduce a novel audit protocol designed to significantly increase the post-audit detectability of such manipulations by enabling the auditor to query the model in an oblivious manner. Our approach leverages a Private Information Retrieval mechanism to require the provider to label a large set of instances, while preventing it from knowing which subset will ultimately be used for the audit. The protocol is efficient, imposes minimal overhead on the auditor, and requires no modification to the audited model, its training procedure, or its inference pipeline. We provide theoretical guarantees showing that, under this protocol, a provider attempting to hide unfairness must falsify a significantly larger number of responses, thereby increasing both the difficulty and the likelihood of detection of manipulation. Experimental results across representative audit scenarios confirm the effectiveness and practicality of our approach. --- ### NeuMoSync: End-to-End Neuromodulatory Control for Plasticity and Adaptability in Continual Learning Authors: Seyed Roozbeh Razavi Rohani, Khashayar Khajavi, Wesley Chung, Mandana Samiei, Mo Chen Published: 2026-08-05 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2608.04358v1 Continual learning (CL) requires models to learn tasks sequentially, yet deep neural networks often suffer from plasticity loss and poor knowledge transfer, which can impede their long-term adaptability. Drawing high-level inspiration from global neuromodulatory mechanisms in the brain, we introduce Neuromodulation and Synchronization (NeuMoSync), a novel architecture that integrates dynamic, neuron-specific modulation into deep neural networks to enhance their adaptability and plasticity. NeuMoSync extends standard neural network architectures with learnable feature vectors for each neuron that track network-wide historical context and with a module operating at a higher level of abstraction. This module synthesizes neuron-specific signals, conditioned on both current inputs and the network's evolving state, to adaptively regulate activation dynamics and synaptic plasticity. Evaluated on diverse CL benchmarks, including memorization (Random Label CIFAR-10 and Random Label MNIST), concept drift (Shuffle CIFAR-10 and Shuffle Mini-ImageNet), class-incremental learning (Class Split ImageNet and Class Split CIFAR-100), and domain-incremental learning (Permuted MNIST), NeuMoSync demonstrates strong performance in retaining plasticity and achieves improvements in both forward and backward adaptation compared with existing methods. Ablation studies validate the necessity of each component, while analysis of the learned modulatory signals reveals interpretable coordination patterns across tasks. Our work underscores the potential of integrating global coordination mechanisms into deep learning systems to advance robust, adaptive continual learning. The code is publicly available at https://github.com/RoozbehRazavi/NeuMoSync. --- ### The Calibration Floor: Format Repair Can Masquerade as Self-Correction at Small-to-Mid Scale Authors: Mingguang Chen, Bo Qu, Licheng Wang Published: 2026-08-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.04355v1 Accuracy changes after language-model self-revision are usually interpreted as changes in reasoning. We show this can fail at the answer-extraction boundary, and test the failure causally rather than only observationally. Across Qwen3.5 (0.8B-9B), Gemma-4-12B, and two frontier models via API (Tencent Hy3, Nvidia Nemotron-3-Ultra-550B) in 29 primary cells plus a frontier arm, we decompose the always-revise accuracy shift into a content margin (both answers parseable) and format-recovery/loss margins (parseability changes). On 12 cells with meaningful unparseable-answer rates, format effects exceed content effects (Wilcoxon p=1.7e-3). To test this causally, we force already-generated reasoning through grammar-constrained decoding so every answer is parseable by construction: across 14 cells this closes a median 71% of the gap between the naive total effect and the content-margin estimate, with two cells converging exactly and a residual on the two largest-effect cells reported rather than dismissed. A clustered model confirms floor-scale (0.8B/2B) models have far higher odds of content-level change and harm than capable-scale models (p<1e-7). Replicating a cited confidence-gating protocol verbatim on Qwen3.5 does not reproduce its reported gain and shows the same near-zero content margin. A frontier check on much larger models shows format-dominance intensifying with scale: content margin is exactly zero in all 5 cells despite total effects up to +0.275, though this arm is lower-powered. The calibration-floor criterion on the content margin reveals a squeeze: floor-scale cells have headroom but insufficient signal, capable-scale cells have signal but little headroom; only one cell is marginally viable, with negligible sealed-holdout gain. Content is a minority share of what the field has measured as self-correction. We release the instrument, code, and derived results. --- ### HyPASE: Hyperbolic Geometry for Parameter-Efficient Speech Emotion Fine-Tuning Framework for Large Audio-Language Models Authors: Tian Jin, Ruikang Zhang, Zefeng Zhao, Ding Luo, Jin Zeng Published: 2026-08-05 Categories: cs.SD, cs.AI arXiv: https://arxiv.org/abs/2608.04351v1 Large Audio-Language Models (LALMs) excel at general speech understanding; however, adapting them to fine-grained tasks like Speech Emotion Recognition (SER) remains a significant bottleneck. Current Parameter-Efficient Fine-Tuning (PEFT) methods typically operate in flat Euclidean space, and this geometry fails to capture the multi-granularity nature of emotion cues, which range from low-level prosody to high-level semantics. To address this, we propose HyPASE, a hyperbolic PEFT framework for LALM-based SER. HyPASE leverages the Poincare ball model, using the hyperbolic radius as an explicit proxy for representational granularity. The framework consists of two core components: a Hyperbolic Geometric Adapter (HGA) for layer-adaptive weight modulation, and an Emotion-aware Multi-capacity Cross-modal Aggregator (EMCA) that compresses multi-scale features into compact audio prefixes. Empirical results on standard benchmarks show that HyPASE outperforms Euclidean PEFT baselines across all metrics on MELD and achieves a notable Unweighted Accuracy gain on IEMOCAP, particularly in class-imbalanced emotion recognition, with the accompanying slight Weighted Accuracy trade-off reflecting hyperbolic space's geometric prioritization of minority-class representations; furthermore, HyPASE achieves robust zero-shot cross-dataset generalization within a constrained parameter budget. By grounding the adaptation process in hyperbolic geometry, HyPASE offers a highly efficient path for LALM fine-tuning. --- ### Poly-OPD: Heterogeneous Multi-Teacher On-Policy Distillation for Capability-Selectable Flow Models Authors: Siming Fu, Haojun Xu, Ruizhe He, Zheming Fu, Hualiang Wang, Jie Huang, Xiaoxiao Ma, Mingchen Zhong, Weihu Huang, Xiaoxuan He, Linjiang Huang, Si Liu Published: 2026-08-05 Categories: cs.CV arXiv: https://arxiv.org/abs/2608.04349v1 Leading open text-to-image models often carry complementary strengths: one may lead on preference-aligned aesthetics while another follows compositional instructions more faithfully. However, differences in their autoencoders and noise schedules make it difficult to transfer these strengths across models. In this paper, we present Poly-OPD, a framework that can consolidate complementary strengths of heterogeneous teachers into a single compact flow-matching student. To bridge the incompatible latent spaces of different teachers, Poly-OPD performs on-policy distillation through a pixel bridge. Each student-generated image is re-encoded by a selected teacher's encoder and refined from a noise level matched by magnitude under the teacher's noise schedule. The resulting target is further matched to the student in frozen DINOv2 space, enabling supervision across incompatible latent spaces. To retain complementary capabilities without cross-teacher interference, Poly-OPD uses a gradient compatibility diagnostic to organize its adapters: attention LoRA modules are shared across teachers, whereas feed-forward adapters remain teacher-specific. During distillation, a gap-aware curriculum devotes more training to compositional categories where the student still falls short of the teacher. As each gap narrows, training shifts toward categories with larger remaining gaps. By distilling FLUX.1-dev and Z-Image into a 2.5B SD3.5-Medium student, Poly-OPD improves GenEval from 67.3 to 73.3, surpassing both larger teachers, and raises DrawBench HPSv3 from 9.34 to 11.35, consolidating both strengths within a switchable model. --- ### iStructTab: Structured Feature Sequencing for Multimodal Learning of Image and Tabular Data Authors: Al Zadid Sultan Bin Habib, Md Younus Ahamed, Prashnna Gyawali, Gianfranco Doretto, Donald A. Adjeroh Published: 2026-08-05 Categories: cs.CV, cs.AI, cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.04348v1 Multimodal learning of images and tabular data is often impaired by ineffective representations, resulting in redundancy, dispersion, and generalization problems. To tackle this challenge, we introduce Graph-Enhanced Descriptor Sequencing (GEDS), a structured feature sequencing algorithm grounded in principles from the Column Permutation Problem (CPP). GEDS refines statistical descriptors of the features through similarity graph-based computations, systematically determining an effective feature sequencing. We incorporate GEDS within an order-aware efficient transformer framework, utilizing order-aware memory tokens that explicitly adhere to the derived feature sequencing via a dedicated loss function. Experimental results across multimodal benchmarks demonstrate that iStructTab effectively minimizes feature dispersion, improving predictive performance and robustness, and highlighting the significance of structured feature sequencing in multimodal learning. --- ### Introspecting Alignment Shifts Beyond Behaviors Implanted Through Fine-Tuning Authors: Kotaro Yoshida, Laura Gomezjurado Gonzalez, Yukinori Yamamoto, Yuji Naraki, Ryotaro Shimizu, Wenya Wang Published: 2026-08-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.04347v2 Fine-tuning enables a source model to acquire desired capabilities and behaviors in a target domain while retaining much of its general-purpose competence. However, this adaptation process can also degrade alignment properties that were present in the source model. Recent work has shown that large language models can be trained using LoRA-based modules known as introspection adapters (IAs) to describe behavioral changes induced by fine-tuning. However, existing studies primarily consider settings in which the model is fine-tuned on datasets explicitly designed to implant a specific behavior and is then asked to explain the implanted behavior. This differs from practical deployment scenarios, where the misalignment is not necessarily deliberately implanted, but arises only as a side effect. To bridge this gap, we formulate a novel problem setting, in which the target of introspection is not necessarily a behavior explicitly implanted through fine-tuning, but rather alignment shifts that may emerge as unintended side effects, and we construct a dataset for this setting. Furthermore, to enhance sensitivity to internal model changes, we propose the Delta-Aware Introspection Adapter (DAIA), a novel mechanism designed to explicitly process both base-model activations and activation differences induced by fine-tuning. Our empirical evaluation shows that introspection learning generalizes to unseen fine-tuned models and safety categories, and that DAIA generally outperforms existing IAs. --- ### Looking in the Mirror: Introspecting Side-Effect Misalignments Induced by Fine-Tuning Authors: Kotaro Yoshida, Laura Gomezjurado Gonzalez, Yukinori Yamamoto, Yuji Naraki, Ryotaro Shimizu, Wenya Wang Published: 2026-08-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.04347v1 Fine-tuning enables a source model to acquire desired capabilities and behaviors in a target domain while retaining much of its general-purpose competence. However, this adaptation process can also degrade alignment properties that were present in the source model. Recent work has shown that large language models can be trained using LoRA-based modules known as introspection adapters (IAs) to describe behavioral changes induced by fine-tuning. However, existing studies primarily consider settings in which the model is fine-tuned on datasets explicitly designed to implant a specific behavior and is then asked to explain the implanted behavior. This differs from practical deployment scenarios, where the central concern is often side-effect misalignment: unintended degradation of alignment caused by fine-tuning on tasks that are not obviously related to safety or alignment. To bridge this gap, we formulate a novel problem setting called \emph{side-effect introspection}, in which the target of introspection is not a behavior explicitly implanted through fine-tuning, but rather alignment shifts that emerge as unintended side effects, and we construct a dataset for this setting. Furthermore, to enhance sensitivity to internal model changes, we propose the Delta-Aware Introspection Adapter (DAIA), a novel mechanism designed to explicitly process both base-model activations and activation differences induced by fine-tuning. Our empirical evaluation shows that introspection learning generalizes to unseen fine-tuned models and safety categories, and that DAIA consistently outperforms existing introspection adapters. --- ### Equitable System-Prompt Selection via Constrained Mixed-Strategy GroupDRO Authors: Mengyu Xu, Qiaoxin Yang, Zhihan Liu, Ruiyao Xu, Zachary Liu, Kezhen Chen, Chongyang Gao Published: 2026-08-05 Categories: cs.CL, cs.AI, cs.CE, stat.ML arXiv: https://arxiv.org/abs/2608.04339v1 Large language models are increasingly used for information seeking, yet semantically equivalent questions phrased in different ways can receive answers of considerably different quality. System prompts are widely employed to steer response behavior, but they are typically optimized for average-case quality, so some question phrasings may still receive incomplete or low-quality answers. To address this, we formulate a constrained mixed-strategy GroupDRO framework for system-prompt selection. Instead of optimizing the system-prompt text, the framework assigns weights to system prompts in an existing pool to minimize the worst-case information-quality loss across evaluation metrics and groups, while constraining the mean loss to stay close to that of average-based selection. Because pool generation and selection are decoupled, the method applies to any system-prompt pool and can leverage an ensemble of complementary system prompts rather than a single one. Across five LLMs on two bilingual medical and consumer-finance benchmarks, the constrained method reduces the Overall Mean, Worst 25% Mean, and Worst by 13.1%, 13.2%, and 13.7% on average relative to no mitigation while keeping overall quality close to Average selection. Its multi-prompt weights reveal complementarity across metric-group pairs. Code and data are available at https://github.com/Rainxu09/equitable-system-prompt-selection. --- ### COMPAS: Difficulty-Aware Joint Search for Optimizing Code Generation Authors: Jingzhi Gong, Jie M. Zhang, Gunel Jahangirova, Dong Huang, Mohammad Reza Mousavi, Mark Harman Published: 2026-08-05 Categories: cs.SE, cs.AI arXiv: https://arxiv.org/abs/2608.04336v1 Code generation systems make each LLM call with a model, a prompt, and decoding settings. However, existing optimization methods usually tune only part of these choices or use one fixed configuration for all tasks: global optimizers search one configuration for all tasks, routers choose only a model, and prompt optimizers keep the model and decoding settings fixed. This leaves their joint, group-specific interactions unclear. We therefore examine how these choices interact and observe that prompts and decoding settings interact, tuning effects vary by model, and the best configuration varies by task difficulty. Guided by these observations, we introduce COMPAS (Code-generation Optimization over Models, Prompts, And Decoding Settings), a difficulty-aware method that learns group-specific quality-cost fronts through low-cost model selection and joint prompt-decoding search, then routes each test task to its matching front online without further search. Under a matched search budget on LiveCodeBench, COMPAS improves pass@1 from 45.9% for the best baseline to 52.8% while reducing cost from $36.57 to $4.92. This also transfers to repository-level code generation on SWE-bench, resolving 76.0% of tasks versus 70.0% for the best baseline. Code and the reproducibility artifact are available at https://github.com/gjz78910/COMPAS. --- ### ATLAS: Adaptive Topological Learning with Abstract Successors for Continual Learning Authors: R. Blake Lawlor, Daniel S. Brown Published: 2026-08-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.04334v1 Contemporary model-free reinforcement learning algorithms can achieve very high performance, but have low sample efficiency and are not robust to changes in the environment. Model-based algorithms have much higher sample efficiency, but still fail when the environment shifts. This paper introduces Adaptive Topological Learning with Abstract Successors (ATLAS) to combat these challenges. ATLAS uses a Grow When Required network with Successor Features in order to achieve high sample efficiency while also robustly tackling catastrophic forgetting. We evaluate ATLAS in spatial navigation tasks, benchmarking its performance against common on-policy and off-policy algorithms. Our empirical results demonstrate that by structurally decoupling transition dynamics from the reward signal, ATLAS achieves near-instantaneous adaptation to new goals and can exhibit positive backward transfer, significantly outperforming baseline methods in non-stationary environments. --- ### Cost-Aware Multi-Objective Bandits: Theory and Application to Budgeted LLM Configuration Evaluation Authors: Bo Xue, Zhi Hong, Jiayi Li, Yuanyu Wan, Ji Cheng, Shuang Qiu Published: 2026-08-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.04333v1 Large language model (LLM) configuration evaluation is challenging due to limited evaluation budgets, varying costs, and multiple competing objectives. In this paper, we formulate LLM configuration evaluation as a cost-aware multi-objective bandit problem, where each configuration evaluation incurs a configuration-dependent cost and yields a noisy vector-valued outcome. Under this framework, we study two fundamental problems: online configuration selection and Pareto configuration identification. For online configuration selection, we propose a hypervolume-based UCB algorithm that optimizes an optimistic hypervolume-per-cost index. We establish a budgeted regret bound of order $O\bigl(\sum_{i\ne i^\star}\frac{\log B}{Δ_i}\bigr)$, where $B$ is the evaluation budget, $i^\star$ is the optimal configuration in terms of hypervolume efficiency, and $Δ_i$ is the corresponding efficiency gap of configuration $i$. This bound retains the logarithmic budget dependence of classical single-objective budgeted bandits. For fixed-budget Pareto identification, we develop a cost-aware empirical gap elimination algorithm and prove that its error probability is of order $O\bigl(\exp(-\frac{B}{H_{μ,c}})\bigr)$, where $H_{μ,c}$ is a cost-aware Pareto identification complexity depending on configuration costs and Pareto classification gaps. This error probability decays exponentially with the evaluation budget and recovers the standard Pareto set identification guarantee when all configuration costs are identical. Experiments on LLM configuration evaluation tasks demonstrate that the proposed framework enables efficient online decision-making and accurate cost-aware Pareto identification under limited budgets. --- ### Right Reset: Chunking by Prefix Removal Authors: Mike Vegeto Published: 2026-08-05 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2608.04330v1 Removing the left context from a causal language model reveals a useful kind of boundary: an edge where the model processes the same right-hand tokens with little change. We turn this observation into prefix-removal probing and introduce Right Reset (RR), which measures preservation of the right-hand hidden-state trajectory. A dynamic program converts RR edge scores into variable-length chunks. On flattened text formed by concatenating topically similar records after deleting their separators and layout, RR recovers 47.7% of the original records as clean units, versus 25.9% for a BGE embedding boundary, the strongest tested conventional baseline without task-specific model training. The gain persists after rendering and OCR. Passive scores from the same Qwen3-4B layer and direct prompting of a same-scale instruction model perform substantially worse on flattened records. Across six language models, RR-selected cuts also undergo consistently less local output disruption than unselected candidate edges. An observed-token likelihood-ratio readout is competitive in some architectures, indicating that the central contribution is the intervention: context dependence itself can provide a boundary signal when surface structure is weak. --- ### An Analysis and Implementation of Seam Carving for Content-Aware Image Resizing Authors: Francesco Tosoni Published: 2026-08-05 Categories: cs.CV, cs.GR arXiv: https://arxiv.org/abs/2608.04329v1 Seam carving is a classical content-aware image resizing operator that modifies the width or height of an image by repeatedly removing (or inserting) seams, i.e., 8-connected monotonic paths of pixels of locally minimal importance. Because seams bend around salient content rather than uniformly scaling or cropping it, the operator preserves vital image structures while discarding (or duplicating) low-energy regions. This article describes a C++ implementation of the operator that follows the original formulation of Avidan and Shamir (2007), including the optional forward-energy criterion subsequently introduced by Rubinstein, Shamir and Avidan (2008). The implementation supports image reduction, image enlargement via ordered seam insertion, multi-pass enlargement for large scale factors, a user-supplied weight mask for object protection and removal, along with dumping of energy maps and visualisation of seams. We detail the algorithm, its parameters and its computational complexity, discuss design choices with respect to the original descriptions, and illustrate the behaviour of the operator on natural images. --- ### Real-time probabilistic tsunami forecasting via generative AI Authors: Yusuke Oishi, Takashi Furumura, Fumihiko Imamura Published: 2026-08-05 Categories: cs.LG arXiv: https://arxiv.org/abs/2608.04327v1 Explicit onshore tsunami inundation forecasting can improve public risk awareness, but deterministically predicted inundation boundaries under highly uncertain conditions, such as near-field tsunamis generated by megathrust earthquakes, may falsely imply safety outside the boundaries. Consequently, current warnings primarily target coastal tsunami height, not onshore inundation. Although machine learning enables instant inundation predictions, they remain deterministic, lacking uncertainty quantification. Here, we develop a probabilistic ensemble model based on a conditional diffusion model (a type of generative AI) that reconciles accuracy with calibration. Validated with the 2011 Tohoku-oki earthquake data, our model faithfully tracks the postearthquake uncertainty decreasing over time while accurately predicting inundation depth and extent. Our framework shows that generative AI can shift tsunami forecasting from determinism to probabilism, providing a foundation for next-generation early warning. --- ### Efficient Online Lexicographic Generalized Low-Rank Matrix Bandits Authors: Bo Xue, Ji Cheng, Haodong Jing, Hongzong Li, Shuang Qiu Published: 2026-08-05 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2608.04324v1 This paper studies generalized low-rank matrix bandits with multiple prioritized objectives. At each round, the learner selects a matrix-valued arm and observes a vector-valued reward, whose components correspond to multiple objectives with different priority levels. Each objective is governed by an objective-specific generalized low-rank matrix model, and the learner evaluates arms according to a lexicographic preference order, prioritizing higher-level objectives before lower-level ones. We propose \textsc{Lexi-LowGLM}, an efficient online algorithm that first estimates objective-specific low-rank subspaces and then performs lexicographic learning in the reduced feature spaces. Unlike existing single-objective algorithms that repeatedly solve a batch generalized linear estimator using all historical observations, \textsc{Lexi-LowGLM} updates each objective-specific estimator via an online Newton step, reducing the estimator-update complexity over $T$ rounds from $O(T^2)$ to $O(T)$. We establish a regret bound of $\widetilde O\left(W_i^{\rm lex}\sqrt{m}\,(d_1+d_2)r\sqrt{T}\right)$ for each objective $i\in(m)$, where $r$ is an upper bound on the ranks of the objective-specific parameter matrices and $W_i^{\rm lex}$ characterizes the lexicographic trade-off effect. This bound depends on the effective low-rank dimension $(d_1+d_2)r$ rather than the ambient dimension $d_1d_2$. Numerical experiments further validate the effectiveness and computational efficiency of the proposed method. --- ### DataRx: Missingness-Aware Sampling for Safer Large Language Model Task-Specific Fine-Tuning Authors: Junbo Zhang, Qianli Zhou, Xinyang Deng, Wen Jiang Published: 2026-08-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.04322v1 Task-specific fine-tuning can improve the performance of large language models (LLMs) on downstream tasks. However, our study reveals that task-specific fine-tuning can also weaken the safety guardrails of aligned LLMs. A widely adopted strategy for preserving safety during fine-tuning is to incorporate safety data. Although previous studies have shown that randomly mixing safety data can alleviate safety degradation, the underlying principle determining why some safety examples are more effective than others still remains unclear. In this paper, we propose DataRx, a missingness-aware sampling method for selecting safety-critical examples. DataRx is based on the hypothesis that a safety sample is more effective when the selected examples provide safety signals that fill the missing parts of LLMs' safety capabilities. DataRx's key insight is leveraging high-dimensional hidden representations rather than discrete tokens to quantify the safety signal gap between the target model's native response and the safety reference response. The results show that, with only 1% additional safety samples from BeaverTails, DataRx reduces the average attack success rate of Llama3-8B-Instruct across seven downstream tasks from 59.23% under random sampling to 13.70%. In addition, DataRx can be combined with the existing safety data synthesis method to further enhance safety defenses during fine-tuning. We hope that DataRx will inspire more data-centric defense research. --- ### Trident : How to Break Deep Reinforcement Learning Cyber Defenses (Agentic) Authors: Ryozo Masukawa, Ian Bryant, Armita Kazeminajafabadi, Sanggeon Yun, Hyunwoo Oh, SungHeon Jeong, Nathaniel D. Bastian, Mahdi Imani, Mohsen Imani Published: 2026-08-05 Categories: cs.CR, cs.AI, cs.LG, cs.MA arXiv: https://arxiv.org/abs/2608.04317v2 Autonomous cyber defense systems based on Deep Reinforcement Learning (DRL) have attracted significant research attention, yet remain evaluated almost exclusively against static, heuristic red agents, leaving their robustness against adaptive threats critically understudied. Meanwhile, recent advances in Reinforcement Learning with Verifiable Rewards (RLVR) have improved LLM reasoning, but their integration into cybersecurity remains elusive due to the absence of suitable benchmark environments and interaction datasets. To bridge this gap, we introduce Trident, an agentic LLM red teaming framework comprising three components: a dynamic benchmark with isolated sandbox servers spanning CybORG CAGE 4 and CyberWheel, a dataset comprises over 13,000 high-fidelity red-blue interaction trajectories for RLVR, and a ``Code-as-Policy'' RLVR agentic architecture Trident Agentic). The latter reformulates red agent training as a contextual bandit via a tripartite Log Summarizer--Planner--Coder design, where a trainable Planner generates complete attack strategies from compressed execution logs, which a frozen Coder translates into executable Python policies deployed against live DRL defenders. Empirical evaluations reveal a fundamental brittleness in existing defenses: with a single trainable 7B planner, Trident reduces blue agent defensive performance by an average of 522% compared to static red agent baselines while autonomously discovering emergent behaviors such as decoy avoidance and adaptive state prioritization that static heuristics entirely fail to uncover. --- ### Trident : How to Break Deep Reinforcement Learning Cyber Defenses (Agentic) Authors: Ryozo Masukawa, Ian Bryant, Armita Kazeminajafabadi, Sanggeon Yun, Hyunwoo Oh, SungHeon Jeong, Nathaniel D. Bastian, Mahdi Imani, Mohsen Imani Published: 2026-08-05 Categories: cs.CR, cs.AI, cs.LG, cs.MA arXiv: https://arxiv.org/abs/2608.04317v1 Autonomous cyber defense systems based on Deep Reinforcement Learning (DRL) have attracted significant research attention, yet remain evaluated almost exclusively against static, heuristic red agents, leaving their robustness against adaptive threats critically understudied. Meanwhile, recent advances in Reinforcement Learning with Verifiable Rewards (RLVR) have improved LLM reasoning, but their integration into cybersecurity remains elusive due to the absence of suitable benchmark environments and interaction datasets. To bridge this gap, we introduce Trident, an agentic LLM red teaming framework comprising three components: a dynamic benchmark with isolated sandbox servers spanning CybORG CAGE 4 and CyberWheel, a dataset comprises over 13,000 high-fidelity red-blue interaction trajectories for RLVR, and a ``Code-as-Policy'' RLVR agentic architecture Trident Agentic). The latter reformulates red agent training as a contextual bandit via a tripartite Log Summarizer--Planner--Coder design, where a trainable Planner generates complete attack strategies from compressed execution logs, which a frozen Coder translates into executable Python policies deployed against live DRL defenders. Empirical evaluations reveal a fundamental brittleness in existing defenses: with a single trainable 7B planner, Trident reduces blue agent defensive performance by an average of 522% compared to static red agent baselines while autonomously discovering emergent behaviors such as decoy avoidance and adaptive state prioritization that static heuristics entirely fail to uncover. --- ### Adversarial Attacks for Good: A Survey of Proactive Protection across the Visual Content Lifecycle Authors: Jiaming Zhang, Boyang Chen, Zherui Li, Fuyao Zhang, Xinyu Yan, Hong Xi Tae, Wenwen He, Xuan Wang, Siqi Guo, Junhao Dong, Kun Wang, Hanxun Huang et al. Published: 2026-08-05 Categories: cs.CR, cs.CV arXiv: https://arxiv.org/abs/2608.04314v1 Once visual content enters an AI pipeline, its owner often retains little technical control over how it is used. Legal and regulatory remedies can address misuse, but many technical interventions must be applied earlier, when content is released or accessed. This survey examines the protective paradigm that has grown around this intervention point, which we call \emph{adversarial attacks for good}. Perturbations and structured signals long studied as attacks on learned models are instead applied by data owners, creators, platforms, or auditors to disrupt unauthorized automation or support later accountability. Five research communities have arrived at this inversion largely independently, each addressing a different stage of a visual asset's lifecycle: privacy filters against unwanted recognition at sharing time, unlearnable examples against unauthorized training, generative safeguards against malicious editing or imitation, adversarial CAPTCHAs for access control against automated agents, and provenance mechanisms for post-circulation attribution. Although developed in separate venues with incompatible success criteria, many of these methods exploit persistent gaps between human perception, semantic interpretation, and machine inference, suggesting that the paradigm remains relevant as visual pipelines evolve toward multimodal models and autonomous agents. To make their claims comparable, we evaluate all five families along shared axes of transferability, adaptability, and deployment readiness. Across the lifecycle, we find that most protections are still validated mainly against static or weakly adaptive adversaries, while evidence beyond controlled benchmarks remains scarce. We close by consolidating cross-stage countermeasures and open problems for robust, composable, and deployable owner-side protection. --- ### Achieving First-Order Statistical Improvements in Data-Driven Optimization: From No-Free-Lunch to Amplified Decision Perturbation Authors: Henry Lam, Tianyu Wang Published: 2026-08-05 Categories: math.OC, stat.ML arXiv: https://arxiv.org/abs/2608.04312v1 Recent proliferation of data-optimization integration has led to a range of methods that aim to improve the statistical performance of data-driven optimization decisions. However, while many of these methods are motivated intuitively from a robustness or regularization perspective, their resulting statistical benefits are often unclear and, even if available, are established on a case-by-case basis. We provide a systematic dissection of data-driven optimization formulations using the view of "directionally perturbed" empirical optimization (EO). Specifically, this umbrella of formulations, which we call "EO+", covers many existing data-driven optimization methods, including regularization, distributionally robust optimization, transfer learning, and analogous methods for contextual optimization. On the one hand, we argue that without additional, correctly specified, side information, any EO+ method can result in at most second-order improvements. This provides a negative conclusion, namely ``no free lunch is possible", on the statistical power of EO+. On the other hand, we show that when leveraging side information that is geometrically effective, achieving first-order improvements is possible by choosing hyperparameters that are significantly larger than what is typically suggested in the literature. Moreover, we construct a principled methodology based on excess risk estimation, via either system knowledge or bootstrap resampling, to maximize the first-order gain. We demonstrate how this gain connects to the control-variate principle, a variance reduction technique in the Monte Carlo simulation literature, which helps explain why geometrically effective side information is necessary. --- ### Pun Intended: Multi-Agent Translation of Wordplay with Contrastive Learning and Phonetic-Semantic Embeddings Authors: Russell Taylor, Benjamin Herbert, Michael Sana Published: 2026-08-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.04311v1 Translating wordplay across languages has long challenged both professional translators and machine translation systems. We investigate three approaches to translating puns from English to French by combining large language models with linguistic constraints for wordplay generation. Our baseline uses a large language model with feedback from a discriminator prompted with positive and negative French examples. Our guided reasoning pipeline uses combined phonetic-semantic embeddings to retrieve lexical candidates for wordplay generation. Finally, our multi-agent framework iteratively evaluates and regenerates candidate translations using specialized feedback. Moving beyond literal translation, our objective is to preserve the linguistic creativity, ambiguity, and humor of the source-text wordplay rather than simply reproduce its vocabulary. The multi-agent and guided chain-of-thought systems ranked first and second, respectively, in the CLEF JOKER 2025 Task 2 competition under expert human evaluation, despite only modest improvements in BLEU and BERTScore. These findings suggest that both explicit phonetic-semantic guidance and iterative multi-agent evaluation can improve LLM-based wordplay translation relative to direct discriminator-guided generation, particularly when balancing semantic fidelity, phonetic similarity, and natural target-language expression --- ### ArborEnum: Decision Tree Rashomon Sets over Continuous Features Authors: Zakk Heile, Hayden McTavish, Margo Seltzer, Cynthia Rudin Published: 2026-08-05 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.04310v3 The Rashomon effect describes the phenomenon that many models can achieve nearly equivalent performance on the same learning task, with significant ramifications for robustness, feature importance, and customizability. These use cases motivate the computation of Rashomon sets: the set of all models whose regularized loss is near-optimal. Decision trees are one of the few model classes for which Rashomon sets can be fully enumerated, but this computation has always been conditional on a binarization of the original data, either restricting which splits each tree is allowed to make or substantially increasing the complexity of an already difficult combinatorial problem. We introduce the first algorithm that exactly enumerates decision-tree Rashomon sets while exploiting the ordered structure of continuous features. We further develop a relaxation for approximate enumeration and an anytime algorithm that progressively refines the set of candidate thresholds, producing increasingly detailed approximations that converge to the continuous-feature Rashomon set. Experiments show that coarse binarization can miss many trees, important features, and predictive multiplicity; our algorithms achieve orders-of-magnitude speedups over existing enumeration methods, with approximations providing further speedups while maintaining near-perfect recall. --- ### ArborEnum: Decision Tree Rashomon Sets over Continuous Features Authors: Zakk Heile, Hayden McTavish, Margo Seltzer, Cynthia Rudin Published: 2026-08-05 Categories: cs.LG, stat.ML arXiv: https://arxiv.org/abs/2608.04310v1 The Rashomon effect describes the phenomenon that many models can achieve nearly equivalent performance on the same learning task, with significant ramifications for robustness, feature importance, and customizability. These use cases motivate the computation of Rashomon sets: the set of all models whose regularized loss is near-optimal. Decision trees are one of the few model classes for which Rashomon sets can be fully enumerated, but this computation has always been conditional on a binarization of the original data, either restricting which splits each tree is allowed to make or substantially increasing the complexity of an already difficult combinatorial problem. We introduce the first algorithm that exactly enumerates decision-tree Rashomon sets while exploiting the ordered structure of continuous features. We further develop a relaxation for approximate enumeration and an anytime algorithm that progressively refines the set of candidate thresholds, producing increasingly detailed approximations that converge to the continuous-feature Rashomon set. Experiments show that coarse binarization can miss many trees, important features, and predictive multiplicity; our algorithms achieve orders-of-magnitude speedups over existing enumeration methods, with approximations providing further speedups while maintaining near-perfect recall. --- ### MIDAS: Multi-LLM Iterative Data-Adaptive Summarization Authors: Karen Lee, Dhanashree Balaram, Seojun Shon, Umair Rasheed Published: 2026-08-05 Categories: cs.CL, cs.AI, cs.MA arXiv: https://arxiv.org/abs/2608.04307v1 Text summarization is deceptively difficult. While condensing information seems straightforward, real-world enterprise summarization of support tickets, legal documents, incident reports, and more, demands strict adherence to domain-specific guidelines, output formats, and organizational conventions. Crafting prompts that reliably satisfy these constraints is labor-intensive, requiring significant human expertise and continuous maintenance as requirements evolve. Existing automated prompt optimization methods reduce this burden through Large Language Model (LLM) critique-driven refinement, yet remain limited by static prompts that cannot adapt to the diversity of summary applications. We propose Multi-LLM Iterative Data-Adaptive Summarization (MIDAS), a multi-LLM framework that extends this paradigm with data-driven pattern learning and use-case-specific personalization, enabling automatic adaptation to different summarization requirements without manual prompt engineering. Applied to enterprise customer ticket summarization across five output formats, MIDAS achieves the strongest overall performance against state-of-the-art critique-driven optimization frameworks such as CriSPO and ZERA, improving ROUGE-1 by up to 11.0%, ROUGE-2 by up to 18.2%, and ROUGE-L by up to 8.0%, while consistently improving BERTScore F1 across all formats and output types. We additionally demonstrate cross-model and cross-domain generalization through multi-LLM configurations and finance-domain summarization benchmarks. --- ### Adaptive Finite-Budget Training for CVaR Risk-Aware Q-Learning Authors: Yifan Wu, Junjie Lei, Wenjie Huang Published: 2026-08-05 Categories: cs.LG, q-fin.RM arXiv: https://arxiv.org/abs/2608.04305v1 Risk-aware Q-learning (RaQL) provides a model-free, two-timescale estimator for dynamic risk objectives, but its finite-budget behavior remains fragile: fixed inner-loop hyperparameters can produce unstable value estimates, persistent Bellman residuals, and inefficient sample reuse. This paper proposes an adaptive training controller for Conditional Value-at-Risk (CVaR) RaQL and evaluates it on a daily Bitcoin trading task. The controller preserves the original CVaR estimator and Bellman fixed point; instead, it redesigns the training procedure through six coordinated mechanisms: per-cell inner-step sizing, outer-rate-matched decay synchronization, a short early correction for the VaR-like inner variable, a coverage-first-then-greedy sample allocation rule, progressive suffix aggregation of mature inner estimates, and data-driven calibration of key scales from online-observable quantities. Across 20 random seeds and 856,000 inner-transition samples, the controller reduces the mean empirical CVaR Bellman residual by approximately 85% relative to the fixed-parameter baseline (MeanBEQ: 1.2202 to 0.1854; MeanBEV: 1.1624 to 0.0535) and maintains stability across CVaR levels, discount factors, and training budgets. On the chronological out-of-sample test set, the learned policy attains a Sharpe ratio of 0.9281 with a maximum drawdown of 6.46% after transaction costs. Although buy-and-hold yields a higher cumulative return (35.43% vs. 23.61%), the adaptive policy achieves far lower volatility (9.57% vs. 47.93%), drawdown, and CVaR loss. These results demonstrate that adaptive finite-budget training design, applied solely to the training procedure without altering the risk objective, can materially improve the reliability and risk-adjusted performance of risk-aware Q-learning in financial applications. --- ### CLIP-CC-Bench: Evaluating Paragraph-Level Video Descriptions in Video-Language Models Authors: Mukhtiar Ali, Harsh Dubey, Sugam Mishra, Chulwoo Pack Published: 2026-08-05 Categories: cs.CV, cs.IR, cs.MM arXiv: https://arxiv.org/abs/2608.04302v1 Benchmarking video-language models has largely focused on short clips and single-sentence metrics, leaving open whether current systems can generate accurate long-form, paragraph-level descriptions. We introduce CLIP-CC-Bench, an evaluation suite for long-form video description built from 5 hours of movie content segmented into 90-second clips, each paired with an expert-written paragraph-style reference. The evaluation suite employs an ensemble of five state-of-the-art LLM-based embedding models to increase reliability and mitigate single-model bias, and applies two complementary methodologies: (i) coarse-grained semantic matching and (ii) fine-grained semantic matching to compare model-generated descriptions against CLIP-CC-Bench references. Using this framework, we evaluate 17 state-of-the-art video-language models and report both their Borda-aggregated rankings and their average scores on CLIP-CC-Bench. We further quantify the protocol's internal reliability through inter-judge agreement and bootstrap ranking stability. We release standardized evaluation scripts, model outputs, and aggregation tools at https://github.com/Multimodal-Intelligence-Lab/CLIP-CC-Bench to support reproducibility. CLIP-CC-Bench provides a practical evaluation framework for long-form video description, filling a gap left by existing short-clip and QA-only benchmarks. --- ### Searching for Sound-Meaning Collisions: Graph-Based Affordance Retrieval and Multi-Evaluator Ranking for Pun Translation at CLEF 2026 JOKER Task 2 Authors: Russell Taylor, Adam Brikman, Prateek Awate Published: 2026-08-05 Categories: cs.CL arXiv: https://arxiv.org/abs/2608.04299v1 Fifteen years ago, Low proposed that pun translators should stop searching for equivalent words and instead search for new points of contact between sound and meaning. In this paper, we investigate that idea computationally. We model pun translation as a process of discovery, exploration, and selection. A retrieval system searches semantic and phonological neighborhoods for target-language affordances: sound-meaning bridges that may support new wordplay. Multiple language models then explore these opportunities by generating competing translations, while a multi-perspective generate-and-rank architecture selects among them. Beyond system development, our primary contribution is an analysis of how retrieved affordances propagate through the translation process. We find that generators actively exploit retrieved opportunities, evaluators progressively concentrate around stronger sound-meaning bridges, and exact phonological collisions are selected at disproportionately high rates when available. At the same time, many puns still yield no usable affordances, suggesting that retrieval remains the central bottleneck in computational pun translation. The resulting picture is remarkably close to the process envisioned by Low. Successful pun translation emerges not from preserving source-language words, but from discovering new places in the target language where sound and meaning collide. --- ### CrossScale-GLIO: Topology-Preserving Vision-Language Alignment of MRI and Whole-Slide Histopathology for Diffuse Glioma Authors: Yantong Liu, Zheyu Zhang, Runpeng Liu, Mu Xitang, Seong-Yoon Shin, Hyun-Ae Lee Published: 2026-08-04 Categories: eess.IV, cs.AI, q-bio.QM arXiv: https://arxiv.org/abs/2609.28524v1 Magnetic resonance imaging and histopathology observe the same glioma at radically different scales. We present CrossScale-GLIO, a visual multimodal framework that represents MRI as a tumor-habitat graph and histology as a cell-niche graph, then aligns them with a structure-aware optimal transport objective anchored by diagnostic language. Across paired and external glioma cohorts, CrossScale-GLIO achieved a paired-test subtype macro-F1 of 0.789, IDH AUROC of 0.934, 1p/19q AUROC of 0.884, and MGMT AUROC of 0.802. The subtype gain over feature-only transport was 2.8 percentage points (95% CI: 1.2 to 4.4, adjusted p = 0.0019). Bidirectional patient retrieval reached Recall@1 values of 0.286 and 0.278, and Recall@5 values of 0.621 and 0.608. Pathologists rated 81.2% of high-mass habitat-niche pairs as biologically plausible. Deleting the highest-mass pair reduced correct-class probability by 0.184, compared with 0.049 under random deletion. Degree-preserving graph rewiring reduced subtype macro-F1 by 0.034 and retrieval Recall@1 by 0.090, directly confirming that preserved relational topology drives cross-scale correspondence. --- ### One Domain, Many Tongues: Composing Domain and Language LoRAs for Cross-Lingual Remote-Sensing MLLMs without Paired Data Authors: Xuechen Li Published: 2026-08-04 Categories: cs.CL, cs.CV, cs.LG arXiv: https://arxiv.org/abs/2609.26097v1 Remote-sensing (RS) multimodal large language models (MLLMs) are trained and evaluated only in English, while text-only instruction data covers over 100 languages. We propose MODL (Mutually Orthogonal Domain-Language composition), a recipe that adds new languages to an English RS MLLM without a single multilingual RS example: a domain LoRA trained on English RS imagery and a language LoRA trained on text alone are learned jointly, under one loss term that keeps the two updates mutually orthogonal at every layer throughout training. This constraint is the recipe's active ingredient. Without it, the same training answers RS questions correctly but in English, erases much of the base model's multilingual text ability, and diverges on one seed in three; sixteen alternatives, from training-free merging to prior orthogonality variants, fail the same way. MODL repairs every failure on every seed: answers are correct and in the target language 56-71% of the time, where the best alternative reaches 27% and most stay below 8%, text ability stays at the level of the untrained base, and on Spanish it surpasses Qwen2.5-VL-7B, with zero multilingual-multimodal data. A single five-language adapter retains English, Spanish, and Vietnamese at full strength across three seeds; non-Latin scripts remain an open boundary. --- ### RECAP: Relation Evidence Calibration for Detecting Spatial Relation Hallucinations in Vision-Language Models Authors: Feixiang Liu, Qiang Qiu, Qingyang Li, Hui Xu Published: 2026-08-04 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.26093v1 Vision-language models can answer spatial relation questions confidently even when the image supports an incompatible relation. We formulate relation-grounded selective prediction: accept or reject an already-produced yes/no answer by auditing its visual support, rather than treating uncertainty as evidence. RECAP, our relation-evidence calibration framework, compares image-conditioned likelihoods for a claim, its semantic contradictions, and optional one-sided supports, then converts these witnesses into an answer-conditioned rejection risk. A calibration-only gate preserves confidence as a veto when confidence is demonstrably informative and otherwise deploys relation evidence alone. Across 20 group/image-disjoint splits, RECAP lowers H-FPR@80 over confidence by between 2.0 and 17.9 points on VSR and raises Acc@80 by 3.0, 8.6, and 12.6 points on What'sUp for Qwen3-VL-8B, InternVL3.5-8B, and LLaVA-1.5-7B. It outperforms matched VCD-style visual contrast on all four primary metrics in all six settings. Full-pool VSR fallback, target-ranked GSR-Bench transfer, equal-budget supervised controls, and two additional checkpoints show a consistent operating principle: structured counterevidence complements certainty when confidence is misaligned, while the gate retains confidence when it is already useful. --- ### SpecialEduBench: Benchmarking Vision-Language Models on Knowledge, Skill, and Attitude in Language Intervention for Autistic Children Authors: Jihoi Na, Taeyeong Kim, Sungjune Kong, Jaemin Jung, Min Joung Park, Kyungtae Joo, Ahhyun Kim, Shim Jaechang, Sooyoung Joo, Dongjin Ka, SeJoong Kim, Jimin Kim et al. Published: 2026-08-04 Categories: cs.CL arXiv: https://arxiv.org/abs/2609.26090v1 Language is the target of most early intervention for autistic children. Because the goal and the method change from child to child, the work falls to a teacher who takes one child at a time and judges each scene as it unfolds. Artificial intelligence is now being brought to that work, yet the benchmarks that reach special education ask what a model knows rather than what it does in front of a child. Building one is not straightforward, since whether a response is good teaching depends on what the child has just done, so no answer key applies. The evidence that settles it is visual as much as verbal, since the length of a wait, a shift of gaze, and the child's uptake leave no trace in a transcript. We introduce \emph{SpecialEduBench}, which measures pedagogical competence along knowledge, skill, and attitude, with 4,537 knowledge items and with 200 skill items and 68 attitude items built on recorded intervention, the attitude items crossing pressure with monitoring into 192 response cells. Seven special-education experts wrote, scored, and reviewed the items, and we revised the judge model's instruction against the reference scores they set. Across eight frontier vision-language models no axis is saturated, since the strongest still fails about a tenth of the honesty cells. The models converge where the knowledge is factual and separate where the task is situated, and the failures gather where pressure is applied. We intend the benchmark as an audit to run before deployment and as a starting point for models built for this domain. --- ### BDSLI: A hybrid CNN-Transformer model for Bengali Sign Language interpretation Authors: Abir Bin Yousuf, Muhammad Iqbal Hossain Published: 2026-08-04 Categories: cs.CV arXiv: https://arxiv.org/abs/2609.26088v1 This study introduces a novel hybrid CNN-Transformer architecture to address the limited progress in Bengali SLR, focusing on isolated sign word recognition and sentence generation. This specific model combination is new to Bengali SLR tasks. A custom video dataset was developed, featuring 62 distinct Bengali sign words (250 samples/class), along with a separate test dataset. The CNN-Transformer model demonstrated superior performance against all comparative and baseline models (e.g., CNN-LSTM, standalone TCN), achieving a 99.58% training accuracy (99.48% validation) and a 98.65% test accuracy. The trained model was subsequently deployed in a web application for real-world validation. --- ### Fast Matrix Multiplication in fp8: Certified Coefficient Optimization and Measured Error Authors: Shuxiao Xie, Shuyang Xie, Yuan Cao, Dezhi Ran, Wei Yang, Tao Xie Published: 2026-08-04 Categories: cs.LG, math.NA arXiv: https://arxiv.org/abs/2609.26077v1 A Strassen-type algorithm has many realizations with the same exact product and multiplication count yet different fp8 error because basis changes reshape coefficient geometry, posing the question of which to run. No current account settles this: classical stability controls worst-case $\ell_1$ growth, not the expected-error magnitude, and the Dumas--Pernet--Sedoglavic optimizer could only be called probably optimal, its global optimality unproved. To settle this, we attach to each realization a coefficient functional $Φ$, a scalar summary of its coefficient geometry, which we minimize over the change-of-basis orbit. This Kempf--Ness problem on a Hadamard manifold lets us certify the global $Φ$ optimum rather than merely search for it: an exact moment-map zero fixes $Φ_{\min} = 200/9$, and de Groote's classification extends that optimality to every exact real rank-7 $2\times2$ decomposition. Every exact real rank-7 realization therefore has a $Φ$-predicted RMS constant at least $5/3$ times that of the cubic algorithm, at fixed noise coefficient. We then introduce an explicit block-scaled e4m3 model in which $Φ$ is the leading-order coefficient of relative expected mean-squared error, and we test the resulting $Φ$-predicted ordering against realized fp8 error. Ordering and re-basing experiments support that prediction within tested fused block-scaled regimes, and on real matmul tiles from four architecture families the $Φ$-optimal realization falls in the fp8 low-error region. Across two $\sim$70B models on real deep_gemm kernels, the same realization removes 10 to 55% of classic Strassen's excess NLL over the clean model. Algorithm realization thus becomes a mathematically certified design problem rather than a tuning choice: an independent low-precision axis with a global $Φ$ optimum and measured fp8 relevance. --- ### Selection-Invariant Communication Compilers for Privacy-Aware Multi-Agent LLM Workflows Authors: Jinghan Xu, Longze Fan, Zeyuan Wang, Xinjin Li, Hankai Liu Published: 2026-08-04 Categories: cs.AI, cs.CR arXiv: https://arxiv.org/abs/2609.26076v1 Structured multi-agent workflows exchange intermediate messages whose content and form can reveal private state even when the final output is safe. We identify selection-channel leakage: after authorization fixes what may be released, a private-state-aware choice among semantically valid realizations creates an additional inference channel. We introduce the selection-invariant communication compiler(SICC), which constrains this post-authorization representation kernel rather than prescribing templates. Any deterministic or independently public-randomized generator satisfying the invariant is valid; requirement-indexed canonical forms are one auditable implementation. We prove a compositional communication-layer guarantee: authorization, public-only form generation, and a dependency-safe utility gate make the emitted transcript reveal no information beyond the complete authorized view. Private-state-aware selection remains vulnerable after surface-disjoint and length-matched controls. Across 132 AgentLeak communication replays and 100 executable LangGraph tasks, deterministic SICC retains complete protocol utility without a positive excess-gain signal; independent public randomization preserves the same result in AgentLeak and 480 controlled cases. --- ### FuncCode: Compressing Kolmogorov--Arnold Networks in Function Space with Hardware-Aware Quantization Authors: Kazi Ahmed Asif Fuad, Lizhong Chen Published: 2026-08-04 Categories: cs.LG arXiv: https://arxiv.org/abs/2609.26067v1 Kolmogorov--Arnold Networks (KANs) replace scalar edge weights with learnable univariate functions, increasing flexibility but also parameter memory because each edge stores multiple coefficients, often together with a separate base branch. We introduce FuncCode, a basis-agnostic compression approach that forms shared codebooks from sampled edge responses, codes the basis and base branches independently, and exports the resulting codebooks and per-edge indices in a quantized, bit-packed format. Across spline and polynomial KANs, sampled edge responses exhibit $13$--$35\%$ lower effective rank than their coefficient representations. Further replicated controls show that function-space clustering alone is statistically tied with coefficient-space clustering; the consistent accuracy gain comes from preserving the distinct sharing structure of the two branches. On a ten-seed MNIST benchmark, FuncCode compresses spline and GRAM KANs by $31.6\times$ and $17.6\times$ with only $0.31$ and $0.34$ pp accuracy loss. On a 6.1M-edge convolutional KAGN, it achieves $19.9\times$ compression while remaining within $0.54$ pp of dense accuracy on CIFAR-10 and $1.89$ pp on CIFAR-100. After compression, per-edge indices account for up to $99.4\%$ of stored weight bits, making the representation index-bound. Across nine bit-exact FPGA accelerators, FuncCode reduces SplineKAN post-route weight memory by $3.87\times$ relative to dense INT4, without increasing cycle count or latency. The FuncCode implementation is available at https://github.com/OSU-STARLAB/FuncCode. --- ### VISPATH: Visual-Intent-Guided Path Reasoning for Multimodal Knowledge Graph Question Answering Authors: Jinke Wu, Zhengpin Li, Mengzhe Jia, Yang Li, Wentao Zhang Published: 2026-08-04 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2609.20843v1 Knowledge graph question answering (KGQA) enables models to answer natural-language questions through structured graph reasoning and has achieved substantial progress across many benchmarks and applications. Recently, multimodal KGQA (MM-KGQA) has attracted increasing attention because many questions require jointly using multimodal inputs and KG evidence. However, existing MM-KGQA methods typically use multimodal information only for starting entity grounding or evidence retrieval, after which multi-hop reasoning degenerates into text-only graph search. As a result, they cannot exploit multimodal cues that become important at intermediate hops. To address this limitation, we propose VISPATH, a visual-intent-guided path reasoning framework for MM-KGQA. VISPATH first identifies a reliable starting entity by combining multimodal grounding with graph-structural cues. It then performs intent-guided path discovery by recomputing hop-specific multimodal intent from the input, question, and current partial paths, so that each expansion is guided by the current reasoning state. The discovered paths are further refined through reasoning-chain pruning, which evaluates candidate paths as complete evidence chains based on their consistency with the question, reasoning sketch, and hop-specific intent. Finally, VISPATH checks whether the selected evidence is sufficient for answer generation. We further construct VISPATH-Bench, a benchmark for evaluating multimodal multi-hop reasoning over KGs, covering questions that require two to four hops over KG paths. Extensive experiments on VISPATH-Bench and three additional multimodal QA benchmarks show that VISPATH consistently outperforms strong baselines. Notably, with GPT-4o as the backbone, VISPATH surpasses GPT-5.4 on VISPATH-Bench, achieving a 10.6% relative improvement in average accuracy and a 13.1% improvement at 2-hop reasoning. --- ### COAL-SQL: Coverage-Guided Augmentation and Failure-Driven Learning for Text-to-SQL Post-Training Authors: Qifeng Cai, Xuanguang Pan, Hao Liang, Chang Xu, Wentao Zhang Published: 2026-08-04 Categories: cs.CL, cs.DB arXiv: https://arxiv.org/abs/2609.20842v1 Text-to-SQL translates natural-language questions into executable SQL queries, but open-source large language models still require task-specific post-training for complex, real-world SQL generation. Effective post-training requires both training data that cover the capabilities demanded by the target task and a learning strategy that enables the model to acquire them. Existing datasets provide valuable supervision but incompletely cover SQL structures, while augmentation methods typically expand data without identifying structural gaps. Moreover, supervised fine-tuning (SFT) or reinforcement learning (RL) alone cannot dynamically address weaknesses exposed during training. We propose COAL-SQL, a unified framework combining Coverage-Guided Augmentation (CGA) and Failure-Driven Learning (FDL). CGA uses greedy selection to identify SQL structures missing from the original dataset and constructs complementary examples, improving structural coverage. FDL retains GRPO as the main optimization objective while supplying targeted supervision for unsolved examples. At the step level, it applies SFT to verified reasoning traces generated by a strong LLM for accumulated failures. At the epoch level, it retrieves structurally related examples based on accumulated failures to create targeted practice, helping the model acquire the corresponding SQL capabilities. With only 12,600 distinct post-training examples, COAL-SQL achieves 64.9% execution accuracy on the BIRD development set and outperforms baselines trained at comparable scale. The code is available at https://github.com/TechNomad-ds/COAL-SQL. --- ### JEPA-WAM: Connecting Generated Visual Instructions to World Action Models through JEPA Latent Representations Authors: Tianbin Liu, Jian Zhu, Taiyi Su, Jianjun Zhang, Chong Ma, Zitai Huang, Yi Xu Published: 2026-08-04 Categories: cs.AI, cs.RO arXiv: https://arxiv.org/abs/2609.20277v1 World Action Models (WAMs) have demonstrated strong robotic manipulation capabilities by augmenting pretrained video generative models with action experts. However, current WAMs still show limited instruction-following ability when conditioned solely on text instructions. We argue that this limitation stems in part from a structural imbalance in robot-learning data: rich visual-action trajectories are often paired with sparse and repetitive language annotations, allowing policies to identify tasks from visual context and motion regularities rather than grounding the instruction itself. To address this limitation, we introduce JEPA-WAM, which augments each text instruction with a bank of stochastically generated visual instructions, providing diverse visual cues for instruction following. Specifically, JEPA-WAM uses an off-the-shelf text-to-image generator to sample multiple task-completion images conditioned on the text instruction, without training the generator. Although these generated images may differ from the current visual scene in appearance and layout, they remain semantically aligned with the instruction and serve as visual goal references. To focus on task-level semantics beyond appearance, we encode these references with a frozen V-JEPA 2.1 encoder. The resulting dense goal representations are compressed into compact goal tokens that condition both the video and action experts through cross-attention. We further construct a real-robot instruction-following benchmark covering in-distribution, out-of-distribution scene, and out-of-distribution instruction settings. On this benchmark, JEPA-WAM achieves success rates of 87.3%, 74.5%, and 80.9% in these three settings, outperforming π0 and Fast-WAM by at least 10.0, 27.3, and 14.5 percentage points, respectively. ---