# Research Papers — 2026-04 · bundle page 164 of 233 > Full-text markdown of papers #8151–8200 of 11612 published in 2026-04, newest first. AI-readable mirrors of the full 11612-paper 2026-04 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-04/page/164 [← Previous bundle](https://aipod.dev/papers/archive/2026-04/page/163.md) [Next bundle →](https://aipod.dev/papers/archive/2026-04/page/165.md) ### Securing Retrieval-Augmented Generation: A Taxonomy of Attacks, Defenses, and Future Directions Authors: Yuming Xu, Mingtao Zhang, Zhuohan Ge, Haoyang Li, Nicole Hu, Yongqi Zhang, Zhiyuan Wen, Jason Chen Zhang, Qing Li, Lei Chen Published: 2026-04-09 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2604.08304v3 Retrieval-augmented generation (RAG) extends large language models (LLMs) with external knowledge, but this access path also introduces security risks that existing work often conflates with inherent LLM flaws. We frame secure RAG as securing external knowledge access and organize the literature with SLOT, a taxonomy along four axes: the attack Surface (S) where an adversary acts, the defense Layer (L) that controls the same point, the Objective (O) it breaks following the CIA properties, and the Target (T) it pursues, from a single known query (T1) to target-claim manipulation across a query distribution (T2). Mapping attacks, defenses, remediation, and evaluation onto a six-stage knowledge-access pipeline, we expose two structural mismatches. Finally, we discuss directions for more realistic targets, no-blind-spot and adaptively evaluated defenses, stronger confidentiality, and evaluation for multimodal and agentic RAG. The curated paper list for RAG security is in: https://github.com/TreeAI-Lab/Awesome-RAG-Security. --- ### DMax: Aggressive Parallel Decoding for dLLMs Authors: Zigeng Chen, Gongfan Fang, Xinyin Ma, Ruonan Yu, Xinchao Wang Published: 2026-04-09 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2604.08302v3 We present DMax, a new paradigm for efficient diffusion language models (dLLMs). It mitigates error accumulation in parallel decoding, enabling aggressive decoding parallelism while preserving generation quality. Unlike conventional masked dLLMs that decode through a binary mask-to-token transition, DMax reformulates decoding as a progressive self-refinement from mask embeddings to token embeddings. At the core of our approach is On-Policy Uniform Training, a novel training strategy that efficiently unifies masked and uniform dLLMs, equipping the model to recover clean tokens from both masked inputs and its own erroneous predictions. Building on this foundation, we further propose Soft Parallel Decoding. We represent each intermediate decoding state as an interpolation between the predicted token embedding and the mask embedding, enabling iterative self-revising in embedding space. Extensive experiments across a variety of benchmarks demonstrate the effectiveness of DMax. Compared with the original LLaDA-2.0-mini, our method improves TPF on GSM8K from 2.04 to 5.47 while preserving accuracy. On MBPP, it increases TPF from 2.71 to 5.86 while maintaining comparable performance. On two H200 GPUs, our model achieves an average of 1,338 TPS at batch size 1. Code is available at: https://github.com/czg1225/DMax --- ### GroundingAnomaly: Spatially-Grounded Diffusion for Few-Shot Anomaly Synthesis Authors: Yishen Liu, Hongcang Chen, Pengcheng Zhao, Yunfan Bao, Yuxi Tian, Jieming Zhang, Hao Chen, Zheng Zhi, Yongchun Liu, Ying Li, Dongpu Cao Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08301v1 The performance of visual anomaly inspection in industrial quality control is often constrained by the scarcity of real anomalous samples. Consequently, anomaly synthesis techniques have been developed to enlarge training sets and enhance downstream inspection. However, existing methods either suffer from poor integration caused by inpainting or fail to provide accurate masks. To address these limitations, we propose GroundingAnomaly, a novel few-shot anomaly image generation framework. Our framework introduces a Spatial Conditioning Module that leverages per-pixel semantic maps to enable precise spatial control over the synthesized anomalies. Furthermore, a Gated Self-Attention Module is designed to inject conditioning tokens into a frozen U-Net via gated attention layers. This carefully preserves pretrained priors while ensuring stable few-shot adaptation. Extensive evaluations on the MVTec AD and VisA datasets demonstrate that GroundingAnomaly generates high-quality anomalies and achieves state-of-the-art performance across multiple downstream tasks, including anomaly detection, segmentation, and instance-level detection. --- ### SeLaR: Selective Latent Reasoning in Large Language Models Authors: Renyu Fu, Guibo Luo Published: 2026-04-09 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2604.08299v2 Chain-of-Thought (CoT) has become a cornerstone of reasoning in large language models, yet its effectiveness is constrained by the limited expressiveness of discrete token sampling. Recent latent reasoning approaches attempt to alleviate this limitation by replacing discrete tokens with soft embeddings (probability-weighted mixtures of token embeddings) or hidden states, but they commonly suffer from two issues: (1) global activation injects perturbations into high-confidence steps, impairing reasoning stability; and (2) soft embeddings quickly collapse toward the highest-probability token, limiting exploration of alternative trajectories. To address these challenges, we propose SeLaR (Selective Latent Reasoning), a lightweight and training-free framework. SeLaR introduces an entropy-gated mechanism that activates soft embeddings only at low-confidence steps, while preserving discrete decoding at high-confidence steps. Additionally, we propose an entropy-aware contrastive regularization that pushes soft embeddings away from the dominant (highest-probability) token's direction, encouraging sustained exploration of multiple latent reasoning paths. Experiments on five reasoning benchmarks demonstrate that SeLaR consistently outperforms standard CoT and state-of-the-art training-free methods. --- ### Robust Multi-Objective Optimization for Bicycle Rebalancing in Shared Mobility Systems Authors: Diego Daniel Pedroza-Perez, Gabriel Luque, Sergio Nesmachnow, Jamal Toutouh Published: 2026-04-09 Categories: cs.NE arXiv: https://arxiv.org/abs/2604.08296v1 Dock-based bike-sharing systems exhibit spatial imbalances between bicycle supply and user demand, often addressed through overnight truck-based rebalancing. This work studies static overnight rebalancing under demand uncertainty modeled as a tri-objective optimization problem. The objectives minimize total travel distance, expected unmet demand, and a robustness-oriented unmet demand measure over high-demand scenarios. Route plans are evaluated via a recourse simulation that enforces truck loads and station capacity constraints across multiple demand realizations. The robustness objective supports selecting plans that reduce peak-demand service degradation. Trade-off solutions are approximated with Non-dominated Sorting Genetic Algorithm II using a permutation--partition encoding and domain-specific relocation operators, including a biased best-improvement move for station relocation. Experiments on the real Barcelona Bicing system with 460 stations show well-distributed Pareto sets and substantial contributions to the reference non-dominated set. Greedy constructive baselines mainly yield extreme solutions and are often dominated. --- ### U-CECE: A Universal Multi-Resolution Framework for Conceptual Counterfactual Explanations Authors: Angeliki Dimitriou, Nikolaos Chaidos, Maria Lymperaiou, Giorgos Filandrianos, Giorgos Stamou Published: 2026-04-09 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2604.08295v3 As AI models grow more complex, explainability is essential for building trust, yet concept-based counterfactual methods still face a trade-off between expressivity and efficiency. Representing underlying concepts as atomic sets is fast but misses relational context, whereas full graph representations are more faithful but require solving the NP-hard Graph Edit Distance (GED) problem. We propose U-CECE, a unified, model-agnostic multi-resolution framework for conceptual counterfactual explanations that adapts to data regime and compute budget. U-CECE spans three levels of expressivity: atomic concepts for broad explanations, relational sets-of-sets for simple interactions, and structural graphs for full semantic structure. At the structural level, both a precision-oriented transductive mode based on supervised Graph Neural Networks (GNNs) and a scalable inductive mode based on unsupervised graph autoencoders (GAEs) are supported. Experiments on the structurally divergent CUB and Visual Genome datasets characterize the efficiency-expressivity trade-off across levels, while human surveys and LVLM-based evaluation show that the retrieved structural counterfactuals are semantically equivalent to, and often preferred over, exact GED-based ground-truth explanations. --- ### Can Vision Language Models Judge Action Quality? An Empirical Evaluation Authors: Miguel Monte e Freitas, Rui Henriques, Ricardo Rei, Pedro Henrique Martins Published: 2026-04-09 Categories: cs.CV, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2604.08294v1 Action Quality Assessment (AQA) has broad applications in physical therapy, sports coaching, and competitive judging. Although Vision Language Models (VLMs) hold considerable promise for AQA, their actual performance in this domain remains largely uncharacterised. We present a comprehensive evaluation of state-of-the-art VLMs across activity domains (e.g. fitness, figure skating, diving), tasks, representations, and prompting strategies. Baseline results reveal that Gemini 3.1 Pro, Qwen3-VL and InternVL3.5 models perform only marginally above random chance, and although strategies such as incorporation of skeleton information, grounding instructions, reasoning structures and in-context learning lead to isolated gains, none is consistently effective. Analysis of prediction distributions uncovers two systematic biases: a tendency to predict correct execution regardless of visual evidence, and a sensitivity to superficial linguistic framing. Reformulating tasks contrastively to mitigate these biases yields minimal improvement, suggesting that the models' limitations go beyond these biases, pointing to a fundamental difficulty with fine-grained movement quality assessment. Our findings establish a rigorous baseline for future VLM-based AQA research and provide an actionable outline for failure modes requiring mitigation prior to reliable real-world deployment. --- ### CIAO - Code In Architecture Out - Automated Software Architecture Documentation with Large Language Models Authors: Marco De Luca, Tiziano Santilli, Domenico Amalfitano, Anna Rita Fasolino, Patrizio Pelliccione Published: 2026-04-09 Categories: cs.SE, cs.AI, cs.ET arXiv: https://arxiv.org/abs/2604.08293v1 Software architecture documentation is essential for system comprehension, yet it is often unavailable or incomplete. While recent LLM-based techniques can generate documentation from code, they typically address local artifacts rather than producing coherent, system-level architectural descriptions. This paper presents a structured process for automatically generating system-level architectural documentation directly from GitHub repositories using Large Language Models. The process, called CIAO (Code In Architecture Out), defines an LLM-based workflow that takes a repository as input and produces system-level architectural documentation following a template derived from ISO/IEC/IEEE 42010, SEI Views \& Beyond, and the C4 model. The resulting documentation can be directly added to the target repository. We evaluated the process through a study with 22 developers, each reviewing the documentation generated for a repository they had contributed to. The evaluation shows that developers generally perceive the produced documentation as valuable, comprehensible, and broadly accurate with respect to the source code, while also highlighting limitations in diagram quality, high-level context modeling, and deployment views. We also assessed the operational cost of the process, finding that generating a complete architectural document requires only a few minutes and is inexpensive to run. Overall, the results indicate that a structured, standards-oriented approach can effectively guide LLMs in producing system-level architectural documentation that is both usable and cost-effective. --- ### CAMotion: A High-Quality Benchmark for Camouflaged Moving Object Detection in the Wild Authors: Siyuan Yao, Hao Sun, Ruiqi Yu, Xiwei Jiang, Wenqi Ren, Xiaochun Cao Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08287v2 Discovering camouflaged objects is a challenging task in computer vision due to the high similarity between camouflaged objects and their surroundings. While the problem of camouflaged object detection over sequential video frames has received increasing attention, the scale and diversity of existing video camouflaged object detection (VCOD) datasets are greatly limited, which hinders the deeper analysis and broader evaluation of recent deep learning-based algorithms with data-hungry training strategy. To break this bottleneck, in this paper, we construct CAMotion, a high-quality benchmark covers a wide range of species for camouflaged moving object detection in the wild. CAMotion comprises various sequences with multiple challenging attributes such as uncertain edge, occlusion, motion blur, and shape complexity, etc. The sequence annotation details and statistical distribution are presented from various perspectives, allowing CAMotion to provide in-depth analyses on the camouflaged object's motion characteristics in different challenging scenarios. Additionally, we conduct a comprehensive evaluation of existing SOTA models on CAMotion, and discuss the major challenges in VCOD task. The benchmark is available at https://www.camotion.focuslab.net.cn, we hope that our CAMotion can lead to further advancements in the research community. --- ### Distributed Multi-Layer Editing for Rule-Level Knowledge in Large Language Models Authors: Yating Wang, Wenting Zhao, Yaqi Zhao, Yongshun Gong, Yilong Yin, Haoliang Sun Published: 2026-04-09 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2604.08284v1 Large language models store not only isolated facts but also rules that support reasoning across symbolic expressions, natural language explanations, and concrete instances. Yet most model editing methods are built for fact-level knowledge, assuming that a target edit can be achieved through a localized intervention. This assumption does not hold for rule-level knowledge, where a single rule must remain consistent across multiple interdependent forms. We investigate this problem through a mechanistic study of rule-level knowledge editing. To support this study, we extend the RuleEdit benchmark from 80 to 200 manually verified rules spanning mathematics and physics. Fine-grained causal tracing reveals a form-specific organization of rule knowledge in transformer layers: formulas and descriptions are concentrated in earlier layers, while instances are more associated with middle layers. These results suggest that rule knowledge is not uniformly localized, and therefore cannot be reliably edited by a single-layer or contiguous-block intervention. Based on this insight, we propose Distributed Multi-Layer Editing (DMLE), which applies a shared early-layer update to formulas and descriptions and a separate middle-layer update to instances. While remaining competitive on standard editing metrics, DMLE achieves substantially stronger rule-level editing performance. On average, it improves instance portability and rule understanding by 13.91 and 50.19 percentage points, respectively, over the strongest baseline across GPT-J-6B, Qwen2.5-7B, Qwen2-7B, and LLaMA-3-8B. The code is available at https://github.com/Pepper66/DMLE. --- ### Revisiting Radar Perception With Spectral Point Clouds Authors: Hamza Alsharif, Jing Gu, Pavol Jancura, Satish Ravindran, Gijs Dubbelman Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08282v1 Radar perception models are trained with different inputs, from range-Doppler spectra to sparse point clouds. Dense spectra are assumed to outperform sparse point clouds, yet they can vary considerably across sensors and configurations, which hinders transfer. In this paper, we provide alternatives for incorporating spectral information into radar point clouds and show that, point clouds need not underperform compared to spectra. We introduce the spectral point cloud paradigm, where point clouds are treated as sparse, compressed representations of the radar spectra, and argue that, when enriched with spectral information, they serve as strong candidates for a unified input representation that is more robust against sensor-specific differences. We develop an experimental framework that compares spectral point cloud (PC) models at varying densities against a dense range-Doppler (RD) benchmark, and report the density levels where the PC configurations meet the performance of the RD benchmark. Furthermore, we experiment with two basic spectral enrichment approaches, that inject additional target-relevant information into the point clouds. Contrary to the common belief that the dense RD approach is superior, we show that point clouds can do just as well, and can surpass the RD benchmark when enrichment is applied. Spectral point clouds can therefore serve as strong candidates for unified radar perception, paving the way for future radar foundation models. --- ### When to Trust Tools? Adaptive Tool Trust Calibration For Tool-Integrated Math Reasoning Authors: Ruotao Xu, Yixin Ji, Yu Luo, Jinpeng Li, Dong Li, Peifeng Li, Juntao Li, Min Zhang Published: 2026-04-09 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.08281v2 Large reasoning models (LRMs) have achieved strong performance enhancement through scaling test time computation, but due to the inherent limitations of the underlying language models, they still have shortcomings in tasks that require precise computation and extensive knowledge reserves. Tool-Integrated Reasoning (TIR) has emerged as a promising paradigm that incorporates tool call and execution within the reasoning trajectory. Although recent works have released some powerful open-source TIR models, our analysis reveals that these models still suffer from critical deficiencies. We find that when the reasoning of the model conflicts with the tool results, the model tends to believe in its own reasoning. And there are cases where the tool results are correct but are ignored by the model, resulting in incorrect answers, which we define as "Tool Ignored''. This indicates that the model does not know when to trust or ignore the tool. To overcome these limitations, We introduce Adaptive Tool Trust Calibration (ATTC), a novel framework that guides the model to adaptively choose to trust or ignore the tool results based on the confidence score of generated code blocks. The experimental results from various open-source TIR models of different sizes and across multiple datasets demonstrate that ATTC effectively reduces the "Tool Ignored" issue, resulting in a performance increase of 4.1% to 7.5%. --- ### QARIMA: A Quantum Approach To Classical Time Series Analysis Authors: Nishikanta Mohanty, Bikash K. Behera, Badshah Mukherjee, Pravat Dash, Giuseppe Sergioli, Roberto Giuntini Published: 2026-04-09 Categories: quant-ph, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2604.08277v3 We present QARIMA, a quantum state-similarity-based reconstruction of the classical ARIMA modelling pipeline. Rather than using a quantum circuit as a standalone forecaster, QARIMA preserves ARIMA's interpretable forecasting structure while reformulating its core building blocks through analogous quantum-compatible modules. The framework integrates quantum differencing assessment, QACF/QPACF lag discovery, compact-swap-test state projection, swap-test/VQC-based AR and MA coefficient estimation, and weak-lag refinement within a single ARIMA forecasting workflow. QACF and QPACF serve the functional roles of ACF and PACF for MA and AR lag discovery, but construct lag relevance through quantum measurement geometry rather than direct classical correlation. Given screened candidate orders $(p,d,q)$, AR and MA coefficients are estimated through state-alignment losses incorporating cosine alignment, entropy regularization, phase correction, and norm control. We evaluate QARIMA across environmental, climatic, industrial, and weather time-series datasets using rolling-origin out-of-sample testing against automated classical ARIMA baselines, with performance assessed through MSE, MAPE, and Diebold--Mariano tests. The results show that quantum state-similarity modules can produce competitive and, in several cases, improved forecasting behaviour while preserving ARIMA's transparency, modularity, and interpretability. QARIMA therefore establishes a distinct pathway for quantum-enhanced statistical forecasting: its modules remain functionally analogous to classical ARIMA subroutines, but are not algebraic replicas; they serve the same modelling roles through state overlap, projection, and measurement-driven parameter estimation. --- ### ACF: A Collaborative Framework for Agent Covert Communication under Cognitive Asymmetry Authors: Wansheng Wu, Kaibo Huang, Yukun Wei, Zhongliang Yang, Linna Zhou Published: 2026-04-09 Categories: cs.AI, cs.CR arXiv: https://arxiv.org/abs/2604.08276v1 As generative artificial intelligence evolves, autonomous agent networks present a powerful paradigm for interactive covert communication. However, because agents dynamically update internal memories via environmental interactions, existing methods face a critical structural vulnerability: cognitive asymmetry. Conventional approaches demand strict cognitive symmetry, requiring identical sequence prefixes between the encoder and decoder. In dynamic deployments, inevitable prefix discrepancies destroy synchronization, inducing severe channel degradation. To address this core challenge of cognitive asymmetry, we propose the Asymmetric Collaborative Framework (ACF), which structurally decouples covert communication from semantic reasoning via orthogonal statistical and cognitive layers. By deploying a prefix-independent decoding paradigm governed by a shared steganographic configuration, ACF eliminates the reliance on cognitive symmetry. Evaluations on realistic memory-augmented workflows demonstrate that under severe cognitive asymmetry, symmetric baselines suffer severe channel degradation, whereas ACF uniquely excels across both semantic fidelity and covert communication. It maintains computational indistinguishability, enabling reliable secret extraction with provable error bounds, and providing robust Effective Information Capacity guarantees for modern agent networks. --- ### Floating or Suggesting Ideas? A Large-Scale Contrastive Analysis of Metaphorical and Literal Verb-Object Constructions Authors: Prisca Piccirilli, Alexander Fraser, Sabine Schulte im Walde Published: 2026-04-09 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.08275v1 Metaphor pervades everyday language, allowing speakers to express abstract concepts via concrete domains. While prior work has studied metaphors cognitively and psycholinguistically, large-scale comparisons with literal language remain limited, especially for near-synonymous expressions. We analyze 297 English verb-object pairs (e.g., float idea vs. suggest idea) in ~2M corpus sentences, examining their contextual usage. Using five NLP tools, we extract 2,293 cognitive and linguistic features capturing affective, lexical, syntactic, and discourse-level properties. We address: (i) whether features differ between metaphorical and literal contexts (cross-pair analysis), and (ii) whether individual VO pairs diverge internally (within-pair analysis). Cross-pair results show literal contexts have higher lexical frequency, cohesion, and structural regularity, while metaphorical contexts show greater affective load, imageability, lexical diversity, and constructional specificity. Within-pair analyses reveal substantial heterogeneity, with most pairs showing non-uniform effects. These results suggest no single, consistent distributional pattern that distinguishes metaphorical from literal usage. Instead, differences are largely construction-specific. Overall, large-scale data combined with diverse features provides a fine-grained understanding of metaphor-literal contrasts in VO usage. --- ### Preventing Overfitting in Deep Image Prior for Hyperspectral Image Denoising Authors: Panagiotis Gkotsis, Athanasios A. Rontogiannis Published: 2026-04-09 Categories: cs.CV, eess.IV arXiv: https://arxiv.org/abs/2604.08272v1 Deep image prior (DIP) is an unsupervised deep learning framework that has been successfully applied to a variety of inverse imaging problems. However, DIP-based methods are inherently prone to overfitting, which leads to performance degradation and necessitates early stopping. In this paper, we propose a method to mitigate overfitting in DIP-based hyperspectral image (HSI) denoising by jointly combining robust data fidelity and explicit sensitivity regularization. The proposed approach employs a Smooth $\ell_1$ data term together with a divergence-based regularization and input optimization during training. Experimental results on real HSIs corrupted by Gaussian, sparse, and stripe noise demonstrate that the proposed method effectively prevents overfitting and achieves superior denoising performance compared to state-of-the-art DIP-based HSI denoising methods. --- ### An Illusion of Unlearning? Assessing Machine Unlearning Through Internal Representations Authors: Yichen Gao, Altay Unal, Akshay Rangamani, Zhihui Zhu Published: 2026-04-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.08271v1 While numerous machine unlearning (MU) methods have recently been developed with promising results in erasing the influence of forgotten data, classes, or concepts, they are also highly vulnerable-for example, simple fine-tuning can inadvertently reintroduce erased concepts. In this paper, we address this contradiction by examining the internal representations of unlearned models, in contrast to prior work that focuses primarily on output-level behavior. Our analysis shows that many state-of-the-art MU methods appear successful mainly due to a misalignment between last-layer features and the classifier, a phenomenon we call feature-classifier misalignment. In fact, hidden features remain highly discriminative, and simple linear probing can recover near-original accuracy. Assuming neural collapse in the original model, we further demonstrate that adjusting only the classifier can achieve negligible forget accuracy while preserving retain accuracy, and we corroborate this with experiments using classifier-only fine-tuning. Motivated by these findings, we propose MU methods based on a class-mean features (CMF) classifier, which explicitly enforces alignment between features and classifiers. Experiments on standard benchmarks show that CMF-based unlearning reduces forgotten information in representations while maintaining high retain accuracy, highlighting the need for faithful representation-level evaluation of MU. --- ### Orion-Lite: Distilling LLM Reasoning into Efficient Vision-Only Driving Models Authors: Jing Gu, Niccolò Cavagnero, Gijs Dubbelman Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08266v1 Leveraging the general world knowledge of Large Language Models (LLMs) holds significant promise for improving the ability of autonomous driving systems to handle rare and complex scenarios. While integrating LLMs into Vision-Language-Action (VLA) models has yielded state-of-the-art performance, their massive parameter counts pose severe challenges for latency-sensitive and energy-efficient deployment. Distilling LLM knowledge into a compact driving model offers a compelling solution to retain these reasoning capabilities while maintaining a manageable computational footprint. Although previous works have demonstrated the efficacy of distillation, these efforts have primarily focused on relatively simple scenarios and open-loop evaluations. Therefore, in this work, we investigate LLM distillation in more complex, interactive scenarios under closed-loop evaluation. We demonstrate that through a combination of latent feature distillation and ground-truth trajectory supervision, an efficient vision-only student model \textbf{Orion-Lite} can even surpass the performance of its massive VLA teacher, ORION. Setting a new state-of-the-art on the rigorous Bench2Drive benchmark, with a Driving Score of 80.6. Ultimately, this reveals that vision-only architectures still possess significant, untapped potential for high-performance reactive planning. --- ### Neural-Symbolic Knowledge Tracing: Injecting Educational Knowledge into Deep Learning for Responsible Learner Modelling Authors: Danial Hooshyar, Gustav Šír, Yeongwook Yang, Tommi Kärkkäinen, Raija Hämäläinen, Ekaterina Krivich, Mutlu Cukurova, Dragan Gašević, Roger Azevedo Published: 2026-04-09 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.08263v1 The growing use of artificial intelligence (AI) in education, particularly large language models (LLMs), has increased interest in intelligent tutoring systems. However, LLMs often show limited adaptivity and struggle to model learners' evolving knowledge over time, highlighting the need for dedicated learner modelling approaches. Although deep knowledge tracing methods achieve strong predictive performance, their opacity and susceptibility to bias can limit alignment with pedagogical principles. To address this, we propose Responsible-DKT, a neural-symbolic deep knowledge tracing approach that integrates symbolic educational knowledge (e.g., mastery and non-mastery rules) into sequential neural models for responsible learner modelling. Experiments on a real-world dataset of students' math interactions show that Responsible-DKT outperforms both a neural-symbolic baseline and a fully data-driven PyTorch DKT model across training settings. The model achieves over 0.80 AUC with only 10% of training data and up to 0.90 AUC, improving performance by up to 13%. It also demonstrates improved temporal reliability, producing lower early- and mid-sequence prediction errors and the lowest prediction inconsistency rates across sequence lengths, indicating that prediction updates remain directionally aligned with observed student responses over time. Furthermore, the neural-symbolic approach offers intrinsic interpretability via a grounded computation graph that exposes the logic behind each prediction, enabling both local and global explanations. It also allows empirical evaluation of pedagogical assumptions, revealing that repeated incorrect responses (non-mastery) strongly influence prediction updates. These results indicate that neural-symbolic approaches enhance both performance and interpretability, mitigate data limitations, and support more responsible, human-centered AI in education. --- ### DBMF: A Dual-Branch Multimodal Framework for Out-of-Distribution Detection Authors: Jiangbei Yue, Darren Treanor, Venkataraman Subramanian, Sharib Ali Published: 2026-04-09 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2604.08261v2 The complex and dynamic real-world clinical environment demands reliable deep learning (DL) systems. Out-of-distribution (OOD) detection plays a critical role in enhancing the reliability and generalizability of DL models when encountering data that deviate from the training distribution, such as unseen disease cases. However, existing OOD detection methods typically rely either on a single visual modality or solely on image-text matching, failing to fully leverage multimodal information. To overcome the challenge, we propose a novel dual-branch multimodal framework by introducing a text-image branch and a vision branch. Our framework fully exploits multimodal representations to identify OOD samples through these two complementary branches. After training, we compute scores from the text-image branch ($S_t$) and vision branch ($S_v$), and integrate them to obtain the final OOD score $S$ that is compared with a threshold for OOD detection. Comprehensive experiments on publicly available endoscopic image datasets demonstrate that our proposed framework is robust across diverse backbones and improves state-of-the-art performance in OOD detection by up to 24.84% --- ### Behavior-Aware Item Modeling via Dynamic Procedural Solution Representations for Knowledge Tracing Authors: Jun Seo, Sangwon Ryu, Heejin Do, Hyounghun Kim, Gary Geunbae Lee Published: 2026-04-09 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2604.08260v1 Knowledge Tracing (KT) aims to predict learners' future performance from past interactions. While recent KT approaches have improved via learning item representations aligned with Knowledge Components, they overlook the procedural dynamics of problem solving. We propose Behavior-Aware Item Modeling (BAIM), a framework that enriches item representations by integrating dynamic procedural solution information. BAIM leverages a reasoning language model to decompose each item's solution into four problem-solving stages (i.e., understand, plan, carry out, and look back), pedagogically grounded in Polya's framework. Specifically, it derives stage-level representations from per-stage embedding trajectories, capturing latent signals beyond surface features. To reflect learner heterogeneity, BAIM adaptively routes these stage-wise representations, introducing a context-conditioned mechanism within a KT backbone, allowing different procedural stages to be emphasized for different learners. Experiments on XES3G5M and NIPS34 show that BAIM consistently outperforms strong pretraining-based baselines, achieving particularly large gains under repeated learner interactions. --- ### HyperMem: Hypergraph Memory for Long-Term Conversations Authors: Juwei Yue, Chuanrui Hu, Jiawei Sheng, Zuyi Zhou, Wenyuan Zhang, Tingwen Liu, Li Guo, Yafeng Deng Published: 2026-04-09 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2604.08256v2 Long-term memory is essential for conversational agents to maintain coherence, track persistent tasks, and provide personalized interactions across extended dialogues. However, existing approaches as Retrieval-Augmented Generation (RAG) and graph-based memory mostly rely on pairwise relations, which can hardly capture high-order associations, i.e., joint dependencies among multiple elements, causing fragmented retrieval. To this end, we propose HyperMem, a hypergraph-based hierarchical memory architecture that explicitly models such associations using hyperedges. Particularly, HyperMem structures memory into three levels: topics, episodes, and facts, and groups related episodes and their facts via hyperedges, unifying scattered content into coherent units. Leveraging this structure, we design a hybrid lexical-semantic index and a coarse-to-fine retrieval strategy, supporting accurate and efficient retrieval of high-order associations. Experiments on the LoCoMo benchmark show that HyperMem achieves state-of-the-art performance with 92.73% LLM-as-a-judge accuracy, demonstrating the effectiveness of HyperMem for long-term conversations. --- ### From Phenomenological Fitting to Endogenous Deduction: A Paradigm Leap via Meta-Principle Physics Architecture Authors: Helong Hu, HongDan Pan, ShuiQing Hu Published: 2026-04-09 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.08245v1 The essence of current neural network architectures is phenomenological fitting: they learn input-output statistical correlations via massive parameters and data, yet lack intrinsic understanding of the fundamental principles governing physical reality. This paper proposes a paradigm leap from pure phenomenological fitting to the fusion of phenomenological fitting and endogenous deduction. By embedding physical meta-principles into neural network architecture, we construct the Meta-Principle Physics Architecture (MPPA). Specifically, MPPA embeds three core meta-principles - Connectivity, Conservation, Periodicity - into its architecture, implemented via three core components: the Gravitator realizes Connectivity via standard causal attention; the Energy Encoder implements Conservation via log-domain energy tracking and delayed compensation; the Periodicity Encoder fulfills Periodicity via FFT-based spectral analysis and delayed modulation. These components collaborate via a learnable independent gating fusion mechanism, forming a complete physical cognition framework of 'local relational connectivity - global conservation constraint - evolutionary periodic law'. Experiments show MPPA achieves significant improvements: physical reasoning (from near zero to 0.436, 0.436 vs 0.000), 2.18x mathematical task improvement (0.330 vs 0.151), 52% logical task gain (0.456 vs 0.300), and 3.69% lower validation perplexity (259.45 vs 269.40), with only 11.8% more parameters (242.40M vs 216.91M). Notably, MPPA shows strong generalization on out-of-distribution physical scenarios, proving the robustness and interpretability of this principle-embedded design. This work establishes a new theoretical foundation and technical path for next-generation AI with physical common sense, causal reasoning, and mathematical rigor. --- ### Self-Debias: Self-correcting for Debiasing Large Language Models Authors: Xuan Feng, Shuai Zhao, Luwei Xiao, Tianlong Gu, Bo An Published: 2026-04-09 Categories: cs.CL arXiv: https://arxiv.org/abs/2604.08243v2 Although Large Language Models (LLMs) demonstrate remarkable reasoning capabilities, inherent social biases often cascade throughout the Chain-of-Thought (CoT) process, leading to continuous "Bias Propagation". Existing debiasing methods primarily focus on static constraints or external interventions, failing to identify and interrupt this propagation once triggered. To address this limitation, we introduce Self-Debias, a progressive framework designed to instill intrinsic self-correction capabilities. Specifically, we reformulate the debiasing process as a strategic resource redistribution problem, treating the model's output probability mass as a limited resource to be reallocated from biased heuristics to unbiased reasoning paths. Unlike standard preference optimization which applies broad penalties, Self-Debias employs a fine-grained trajectory-level objective subject to dynamic debiasing constraints. This enables the model to selectively revise biased reasoning suffixes while preserving valid contextual prefixes. Furthermore, we integrate an online self-improvement mechanism utilizing consistency filtering to autonomously synthesize supervision signals. With merely 20k annotated samples, Self-Debias activates efficient self-correction, achieving superior debiasing performance while preserving general reasoning capabilities without continuous external oversight. --- ### $\oslash$ Source Models Leak What They Shouldn't $\nrightarrow$: Unlearning Zero-Shot Transfer in Domain Adaptation Through Adversarial Optimization Authors: Arnav Devalapally, Poornima Jain, Kartik Srinivas, Vineeth N. Balasubramanian Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08238v1 The increasing adaptation of vision models across domains, such as satellite imagery and medical scans, has raised an emerging privacy risk: models may inadvertently retain and leak sensitive source-domain specific information in the target domain. This creates a compelling use case for machine unlearning to protect the privacy of sensitive source-domain data. Among adaptation techniques, source-free domain adaptation (SFDA) calls for an urgent need for machine unlearning (MU), where the source data itself is protected, yet the source model exposed during adaptation encodes its influence. Our experiments reveal that existing SFDA methods exhibit strong zero-shot performance on source-exclusive classes in the target domain, indicating they inadvertently leak knowledge of these classes into the target domain, even when they are not represented in the target data. We identify and address this risk by proposing an MU setting called SCADA-UL: Unlearning Source-exclusive ClAsses in Domain Adaptation. Existing MU methods do not address this setting as they are not designed to handle data distribution shifts. We propose a new unlearning method, where an adversarially generated forget class sample is unlearned by the model during the domain adaptation process using a novel rescaled labeling strategy and adversarial optimization. We also extend our study to two variants: a continual version of this problem setting and to one where the specific source classes to be forgotten may be unknown. Alongside theoretical interpretations, our comprehensive empirical results show that our method consistently outperforms baselines in the proposed setting while achieving retraining-level unlearning performance on benchmark datasets. Our code is available at https://github.com/D-Arnav/SCADA --- ### HiRO-Nav: Hybrid ReasOning Enables Efficient Embodied Navigation Authors: He Zhao, Yijun Yang, Zichuan Lin, Deheng Ye, Chunyan Miao Published: 2026-04-09 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.08232v1 Embodied navigation agents built upon large reasoning models (LRMs) can handle complex, multimodal environmental input and perform grounded reasoning per step to improve sequential decision-making for long-horizon tasks. However, a critical question remains: \textit{how can the reasoning capabilities of LRMs be harnessed intelligently and efficiently for long-horizon navigation tasks?} In simple scenes, agents are expected to act reflexively, while in complex ones they should engage in deliberate reasoning before acting.To achieve this, we introduce \textbf{H}ybr\textbf{i}d \textbf{R}eas\textbf{O}ning \textbf{Nav}igation (\textbf{HiRO-Nav}) agent, the first kind of agent capable of adaptively determining whether to perform thinking at every step based on its own action entropy. Specifically, by examining how the agent's action entropy evolves over the navigation trajectories, we observed that only a small fraction of actions exhibit high entropy, and these actions often steer the agent toward novel scenes or critical objects. Furthermore, studying the relationship between action entropy and task completion (i.e., Q-value) reveals that improving high-entropy actions contributes more positively to task success.Hence, we propose a tailored training pipeline comprising hybrid supervised fine-tuning as a cold start, followed by online reinforcement learning with the proposed hybrid reasoning strategy to explicitly activate reasoning only for high-entropy actions, significantly reducing computational overhead while improving decision quality. Extensive experiments on the \textsc{CHORES}-$\mathbb{S}$ ObjectNav benchmark showcases that HiRO-Nav achieves a better trade-off between success rates and token efficiency than both dense-thinking and no-thinking baselines. --- ### Generalization Under Scrutiny: Cross-Domain Detection Progresses, Pitfalls, and Persistent Challenges Authors: Saniya M. Deshmukh, Kailash A. Hambarde, Hugo Proença Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08230v1 Object detection models trained on a source domain often exhibit significant performance degradation when deployed in unseen target domains, due to various kinds of variations, such as sensing conditions, environments and data distributions. Hence, regardless the recent breakthrough advances in deep learning-based detection technology, cross-domain object detection (CDOD) remains a critical research area. Moreover, the existing literature remains fragmented, lacking a unified perspective on the structural challenges underlying domain shift and the effectiveness of adaptation strategies. This survey provides a comprehensive and systematic analysis of CDOD. We start upon a problem formulation that highlights the multi-stage nature of object detection under domain shift. Then, we organize the existing methods through a conceptual taxonomy that categorizes approaches based on adaptation paradigms, modeling assumptions, and pipeline components. Furthermore, we analyze how domain shift propagates across detection stages and discuss why adaptation in object detection is inherently more complex than in classification. In addition, we review commonly used datasets, evaluation protocols, and benchmarking practices. Finally, we identify the key challenges and outline promising future research directions. Cohesively, this survey aims to provide a unified framework for understanding CDOD and to guide the development of more robust detection systems. --- ### Grounding Clinical AI Competency in Human Cognition Through the Clinical World Model and Skill-Mix Framework Authors: Seyed Amir Ahmad Safavi-Naini, Elahe Meftah, Josh Mohess, Pooya Mohammadi Kazaj, Georgios Siontis, Zahra Atf, Peter R. Lewis, Mauricio Reyes, Girish Nadkarni, Roland Wiest, Stephan Windecker, Christoph Grani et al. Published: 2026-04-09 Categories: cs.AI, cs.HC, eess.SY arXiv: https://arxiv.org/abs/2604.08226v1 The competency of any intelligent agent is bounded by its formal account of the world in which it operates. Clinical AI lacks such an account. Existing frameworks address evaluation, regulation, or system design in isolation, without a shared model of the clinical world to connect them. We introduce the Clinical World Model, a framework that formalizes care as a tripartite interaction among Patient, Provider, and Ecosystem. To formalize how any agent, whether human or artificial, transforms information into clinical action, we develop parallel decision-making architectures for providers, patients, and AI agents, grounded in validated principles of clinical cognition. The Clinical AI Skill-Mix operationalizes competency through eight dimensions. Five define the clinical competency space (condition, phase, care setting, provider role, and task) and three specify how AI engages human reasoning (assigned authority, agent facing, and anchoring layer). The combinatorial product of these dimensions yields a space of billions of distinct competency coordinates. A central structural implication is that validation within one coordinate provides minimal evidence for performance in another, rendering the competency space irreducible. The framework supplies a common grammar through which clinical AI can be specified, evaluated, and bounded across stakeholders. By making this structure explicit, the Clinical World Model reframes the field's central question from whether AI works to in which competency coordinates reliability has been demonstrated, and for whom. --- ### EditCaption: Human-Refined SFT and HAE-DPO for Image Editing Instruction Synthesis Authors: Xiangyuan Wang, Honghao Cai, Yunhao Bai, Chao Hui, Tianze Zhou, Haohua Chen, Hao Shi, Yuling Wu, Yao Hu, Xu Tang, Yibo Chen, Wei Zhu Published: 2026-04-09 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2604.08213v2 High-quality source-target image pairs with precise editing instructions are essential for instruction-guided image editing, yet constructing such training triplets at scale remains costly. Recent pipelines often rely on vision-language models to synthesize editing instructions automatically, but we find that strong VLMs still struggle to describe visual transformations between image pairs. In particular, they exhibit three recurring failure modes: orientation inconsistency, viewpoint ambiguity, and missing fine-grained attributes. In a human evaluation on 400 image pairs, several open-source VLM baselines produce critical-error rates above 47\%, making many synthesized instructions unsuitable for downstream training. To address this, we propose EditCaption, a two-stage post-training pipeline for image editing instruction synthesis. First, we construct a 100K supervised fine-tuning dataset through GLM-based auto-captioning, EditScore filtering, and human refinement. Second, we collect 10K human-annotated preference pairs, where each rejected instruction is labeled with its primary error type and severity. Based on this dataset, we propose Hardness-Adaptive Error-Aware DPO (HAE-DPO), a task-adapted DPO objective that introduces an adaptive margin based on human-labeled severity, failure-mode type, and reference-model hardness. Experiments across three benchmarks demonstrate that our 235B model with SFT+HAE-DPO achieves state-of-the-art performance among open-source and closed models, scoring 4.720 on Eval-400, 4.672 on HQ-Edit, and 4.651 on ByteMorph-Bench -- surpassing Gemini-3-Pro on all three. Human evaluation confirms critical error rates drop from 47.75\% to 17.50\%, with correct rates improving from 41.75\% to 70.25\%, surpassing Gemini-3-Pro (66.00\%). --- ### Vision-Language Foundation Models for Comprehensive Automated Pavement Condition Assessment Authors: Blessing Agyei Kyem, Joshua Kofi Asamoah, Anthony Dontoh, Armstrong Aboah Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08212v1 General-purpose vision-language models demonstrate strong performance in everyday domains but struggle with specialized technical fields requiring precise terminology, structured reasoning, and adherence to engineering standards. This work addresses whether domain-specific instruction tuning can enable comprehensive pavement condition assessment through vision-language models. PaveInstruct, a dataset containing 278,889 image-instruction-response pairs spanning 32 task types, was created by unifying annotations from nine heterogeneous pavement datasets. PaveGPT, a pavement foundation model trained on this dataset, was evaluated against state-of-the-art vision-language models across perception, understanding, and reasoning tasks. Instruction tuning transformed model capabilities, achieving improvements exceeding 20% in spatial grounding, reasoning, and generation tasks while producing ASTM D6433-compliant outputs. These results enable transportation agencies to deploy unified conversational assessment tools that replace multiple specialized systems, simplifying workflows and reducing technical expertise requirements. The approach establishes a pathway for developing instruction-driven AI systems across infrastructure domains including bridge inspection, railway maintenance, and building condition assessment. --- ### SciFigDetect: A Benchmark for AI-Generated Scientific Figure Detection Authors: You Hu, Chenzhuo Zhao, Changfa Mo, Haotian Liu, Xiaobai Li Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08211v1 Modern multimodal generators can now produce scientific figures at near-publishable quality, creating a new challenge for visual forensics and research integrity. Unlike conventional AI-generated natural images, scientific figures are structured, text-dense, and tightly aligned with scholarly semantics, making them a distinct and difficult detection target. However, existing AI-generated image detection benchmarks and methods are almost entirely developed for open-domain imagery, leaving this setting largely unexplored. We present the first benchmark for AI-generated scientific figure detection. To construct it, we develop an agent-based data pipeline that retrieves licensed source papers, performs multimodal understanding of paper text and figures, builds structured prompts, synthesizes candidate figures, and filters them through a review-driven refinement loop. The resulting benchmark covers multiple figure categories, multiple generation sources and aligned real--synthetic pairs. We benchmark representative detectors under zero-shot, cross-generator, and degraded-image settings. Results show that current methods fail dramatically in zero-shot transfer, exhibit strong generator-specific overfitting, and remain fragile under common post-processing corruptions. These findings reveal a substantial gap between existing AIGI detection capabilities and the emerging distribution of high-quality scientific figures. We hope this benchmark can serve as a foundation for future research on robust and generalizable scientific-figure forensics. The dataset is available at https://github.com/Joyce-yoyo/SciFigDetect. --- ### OmniJigsaw: Enhancing Omni-Modal Reasoning via Modality-Orchestrated Reordering Authors: Yiduo Jia, Muzhi Zhu, Hao Zhong, Mingyu Liu, Yuling Xi, Hao Chen, Bin Qin, Yongjie Yang, Zhenbo Luo, Chunhua Shen Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08209v1 To extend the reinforcement learning post-training paradigm to omni-modal models for concurrently bolstering video-audio understanding and collaborative reasoning, we propose OmniJigsaw, a generic self-supervised framework built upon a temporal reordering proxy task. Centered on the chronological reconstruction of shuffled audio-visual clips, this paradigm strategically orchestrates visual and auditory signals to compel cross-modal integration through three distinct strategies: Joint Modality Integration, Sample-level Modality Selection, and Clip-level Modality Masking. Recognizing that the efficacy of such proxy tasks is fundamentally tied to puzzle quality, we design a two-stage coarse-to-fine data filtering pipeline, which facilitates the efficient adaptation of OmniJigsaw to massive unannotated omni-modal data. Our analysis reveals a ``bi-modal shortcut phenomenon'' in joint modality integration and demonstrates that fine-grained clip-level modality masking mitigates this issue while outperforming sample-level modality selection. Extensive evaluations on 15 benchmarks show substantial gains in video, audio, and collaborative reasoning, validating OmniJigsaw as a scalable paradigm for self-supervised omni-modal learning. --- ### Introducing Echo Networks for Computational Neuroevolution Authors: Christian Kroos, Fabian Küch Published: 2026-04-09 Categories: cs.LG, cs.NE arXiv: https://arxiv.org/abs/2604.08204v1 For applications on the extreme edge, minimal networks of only a few dozen artificial neurons for event detection and classification in discrete time signals would be highly desirable. Feed-forward networks, RNNs, and CNNs evolved through evolutionary algorithms can all be successful in this respect but pose the problem of allowing little systematicity in mutation and recombination if the standard direct genetic encoding of the weights is used (as for instance in the classic NEAT algorithm). We therefore introduce Echo Networks, a type of recurrent network that consists of the connection matrix only, with the source neurons of the synapses represented as rows, destination neurons as columns and weights as entries. There are no layers, and connections between neurons can be bidirectional but are technically all recurrent. Input and output can be arbitrarily assigned to any of the neurons and only use an additional (optional) function in their computational path, e.g., a sigmoid to obtain a binary classification output. We evaluated Echo Networks successfully on the classification of electrocardiography signals but see the most promising potential in their genome representation as a single matrix, allowing matrix computations and factorisations as mutation and recombination operators. --- ### MedVR: Annotation-Free Medical Visual Reasoning via Agentic Reinforcement Learning Authors: Zheng Jiang, Heng Guo, Chengyu Fang, Changchen Xiao, Xinyang Hu, Lifeng Sun, Minfeng Xu Published: 2026-04-09 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2604.08203v1 Medical Vision-Language Models (VLMs) hold immense promise for complex clinical tasks, but their reasoning capabilities are often constrained by text-only paradigms that fail to ground inferences in visual evidence. This limitation not only curtails performance on tasks requiring fine-grained visual analysis but also introduces risks of visual hallucination in safety-critical applications. Thus, we introduce MedVR, a novel reinforcement learning framework that enables annotation-free visual reasoning for medical VLMs. Its core innovation lies in two synergistic mechanisms: Entropy-guided Visual Regrounding (EVR) uses model uncertainty to direct exploration, while Consensus-based Credit Assignment (CCA) distills pseudo-supervision from rollout agreement. Without any human annotations for intermediate steps, MedVR achieves state-of-the-art performance on diverse public medical VQA benchmarks, significantly outperforming existing models. By learning to reason directly with visual evidence, MedVR promotes the robustness and transparency essential for accelerating the clinical deployment of medical AI. --- ### Approximation of the Basset force in the Maxey-Riley-Gatignol equations via universal differential equations Authors: Finn Sommer, Vamika Rathi, Sebastian Goetschel, Daniel Ruprecht Published: 2026-04-09 Categories: cs.LG, math.NA arXiv: https://arxiv.org/abs/2604.08194v2 The Maxey-Riley-Gatignol equations (MaRGE) model the motion of spherical inertial particles in a fluid. They contain the Basset force, an integral term which models history effects due to the formation of wakes and boundary layer effects. This causes the force that acts on a particle to depend on its past trajectory and complicates the numerical solution of MaRGE. Therefore, the Basset force is often neglected, despite substantial evidence that it has both quantitative and qualitative impact on the movement patterns of modelled particles. Using the concept of universal differential equations, we propose an approximation of the history term via neural networks which approximates MaRGE by a system of ordinary differential equations that can be solved with standard numerical solvers like Runge-Kutta methods. --- ### Approximation of the Basset force in the Maxey-Riley-Gatignol equations via universal differential equations Authors: Finn Sommer, Vamika Rathi, Sebastian Goetschel, Daniel Ruprecht Published: 2026-04-09 Categories: cs.LG, math.NA arXiv: https://arxiv.org/abs/2604.08194v1 The Maxey-Riley-Gatignol equations (MaRGE) model the motion of spherical inertial particles in a fluid. They contain the Basset force, an integral term which models history effects due to the formation of wakes and boundary layer effects. This causes the force that acts on a particle to depend on its past trajectory and complicates the numerical solution of MaRGE. Therefore, the Basset force is often neglected, despite substantial evidence that it has both quantitative and qualitative impact on the movement patterns of modelled particles. Using the concept of universal differential equations, we propose an approximation of the history term via neural networks which approximates MaRGE by a system of ordinary differential equations that can be solved with standard numerical solvers like Runge-Kutta methods. --- ### Inside-Out: Measuring Generalization in Vision Transformers Through Inner Workings Authors: Yunxiang Peng, Mengmeng Ma, Ziyu Yao, Xi Peng Published: 2026-04-09 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2604.08192v1 Reliable generalization metrics are fundamental to the evaluation of machine learning models. Especially in high-stakes applications where labeled target data are scarce, evaluation of models' generalization performance under distribution shift is a pressing need. We focus on two practical scenarios: (1) Before deployment, how to select the best model for unlabeled target data? (2) After deployment, how to monitor model performance under distribution shift? The central need in both cases is a reliable and label-free proxy metric. Yet existing proxy metrics, such as model confidence or accuracy-on-the-line, are often unreliable as they only assess model output while ignoring the internal mechanisms that produce them. We address this limitation by introducing a new perspective: using the inner workings of a model, i.e., circuits, as a predictive metric of generalization performance. Leveraging circuit discovery, we extract the causal interactions between internal representations as a circuit, from which we derive two metrics tailored to the two practical scenarios. (1) Before deployment, we introduce Dependency Depth Bias, which measures different models' generalization capability on target data. (2) After deployment, we propose Circuit Shift Score, which predicts a model's generalization under different distribution shifts. Across various tasks, both metrics demonstrate significantly improved correlation with generalization performance, outperforming existing proxies by an average of 13.4\% and 34.1\%, respectively. Our code is available at https://github.com/deep-real/GenCircuit. --- ### Equivariant Efficient Joint Discrete and Continuous MeanFlow for Molecular Graph Generation Authors: Rongjian Xu, Teng Pang, Zhiqiang Dong, Guoqiang Wu Published: 2026-04-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.08189v1 Graph-structured data jointly contain discrete topology and continuous geometry, which poses fundamental challenges for generative modeling due to heterogeneous distributions, incompatible noise dynamics, and the need for equivariant inductive biases. Existing flow-matching approaches for graph generation typically decouple structure from geometry, lack synchronized cross-domain dynamics, and rely on iterative sampling, often resulting in physically inconsistent molecular conformations and slow sampling. To address these limitations, we propose Equivariant MeanFlow (EQUIMF), a unified SE(3)-equivariant generative framework that jointly models discrete and continuous components through synchronized MeanFlow dynamics. EQUIMF introduces a unified time bridge and average-velocity updates with mutual conditioning between structure and geometry, enabling efficient few-step generation while preserving physical consistency. Moreover, we develop a novel discrete MeanFlow formulation with a simple yet effective parameterization to support efficient generation over discrete graph structures. Extensive experiments demonstrate that EQUIMF consistently outperforms prior diffusion and flow-matching methods in generation quality, physical validity, and sampling efficiency. --- ### AT-ADD: All-Type Audio Deepfake Detection Challenge Evaluation Plan Authors: Yuankun Xie, Haonan Cheng, Jiayi Zhou, Xiaoxuan Guo, Tao Wang, Jian Liu, Weiqiang Wang, Ruibo Fu, Xiaopeng Wang, Hengyan Huang, Xiaoying Huang, Long Ye et al. Published: 2026-04-09 Categories: cs.SD, cs.AI arXiv: https://arxiv.org/abs/2604.08184v1 The rapid advancement of Audio Large Language Models (ALLMs) has enabled cost-effective, high-fidelity generation and manipulation of both speech and non-speech audio, including sound effects, singing voices, and music. While these capabilities foster creativity and content production, they also introduce significant security and trust challenges, as realistic audio deepfakes can now be generated and disseminated at scale. Existing audio deepfake detection (ADD) countermeasures (CMs) and benchmarks, however, remain largely speech-centric, often relying on speech-specific artifacts and exhibiting limited robustness to real-world distortions, as well as restricted generalization to heterogeneous audio types and emerging spoofing techniques. To address these gaps, we propose the All-Type Audio Deepfake Detection (AT-ADD) Grand Challenge for ACM Multimedia 2026, designed to bridge controlled academic evaluation with practical multimedia forensics. AT-ADD comprises two tracks: (1) Robust Speech Deepfake Detection, which evaluates detectors under real-world scenarios and against unseen, state-of-the-art speech generation methods; and (2) All-Type Audio Deepfake Detection, which extends detection beyond speech to diverse, unknown audio types and promotes type-agnostic generalization across speech, sound, singing, and music. By providing standardized datasets, rigorous evaluation protocols, and reproducible baselines, AT-ADD aims to accelerate the development of robust and generalizable audio forensic technologies, supporting secure communication, reliable media verification, and responsible governance in an era of pervasive synthetic audio. --- ### Long-Term Embeddings for Balanced Personalization Authors: Andrii Dzhoha, Egor Malykh Published: 2026-04-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.08181v1 Modern transformer-based sequential recommenders excel at capturing short-term intent but often suffer from recency bias, overlooking stable long-term preferences. While extending sequence lengths is an intuitive fix, it is computationally inefficient, and recent interactions tend to dominate the model's attention. We propose Long-Term Embeddings (LTE) as a high-inertia contextual anchor to bridge this gap. We address a critical production challenge: the point-in-time consistency problem caused by infrastructure constraints, as feature stores typically host only a single "live" version of features. This leads to an offline-online mismatch during model deployments and rollbacks, as models are forced to process evolved representations they never saw during training. To resolve this, we introduce an LTE framework that constrains embeddings to a fixed semantic basis of content-based item representations, ensuring cross-version compatibility. Furthermore, we investigate integration strategies for causal language modeling, considering the data leakage issue that occurs when the LTE and the transformer's short-term sequence share a temporal horizon. We evaluate two representations: a heuristic average and an asymmetric autoencoder with a fixed decoder grounded in the semantic basis to enable behavioral fine-tuning while maintaining stability. Online A/B tests on Zalando demonstrate that integrating LTE as a contextual prefix token using a lagged window yields significant uplifts in both user engagement and financial metrics. --- ### Aligning Agents via Planning: A Benchmark for Trajectory-Level Reward Modeling Authors: Jiaxuan Wang, Yulan Hu, Wenjin Yang, Zheng Pan, Xin Li, Lan-Zhe Guo Published: 2026-04-09 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.08178v2 In classical Reinforcement Learning from Human Feedback (RLHF), Reward Models (RMs) serve as the fundamental signal provider for model alignment. As Large Language Models evolve into agentic systems capable of autonomous tool invocation and complex reasoning, the paradigm of reward modeling faces unprecedented challenges -- most notably, the lack of benchmarks specifically designed to assess RM capabilities within tool-integrated environments. To address this gap, we present Plan-RewardBench, a trajectory-level preference benchmark designed to evaluate how well judges distinguish preferred versus distractor agent trajectories in complex tool-using scenarios. Plan-RewardBench covers four representative task families -- (i) Safety Refusal, (ii) Tool-Irrelevance / Unavailability, (iii) Complex Planning, and (iv) Robust Error Recovery -- comprising validated positive trajectories and confusable hard negatives constructed via multi-model natural rollouts, rule-based perturbations, and minimal-edit LLM perturbations. We benchmark representative RMs (generative, discriminative, and LLM-as-Judge) under a unified pairwise protocol, reporting accuracy trends across varying trajectory lengths and task categories. Furthermore, we provide diagnostic analyses of prevalent failure modes. Our results reveal that all three evaluator families face substantial challenges, with performance degrading sharply on long-horizon trajectories, underscoring the necessity for specialized training in agentic, trajectory-level reward modeling. Ultimately, Plan-RewardBench aims to serve as both a practical evaluation suite and a reusable blueprint for constructing agentic planning preference data. --- ### Value-Guidance MeanFlow for Offline Multi-Agent Reinforcement Learning Authors: Teng Pang, Zhiqiang Dong, Yan Zhang, Rongjian Xu, Guoqiang Wu, Yilong Yin Published: 2026-04-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.08174v1 Offline multi-agent reinforcement learning (MARL) aims to learn the optimal joint policy from pre-collected datasets, requiring a trade-off between maximizing global returns and mitigating distribution shift from offline data. Recent studies use diffusion or flow generative models to capture complex joint policy behaviors among agents; however, they typically rely on multi-step iterative sampling, thereby reducing training and inference efficiency. Although further research improves sampling efficiency through methods like distillation, it remains sensitive to the behavior regularization coefficient. To address the above-mentioned issues, we propose Value Guidance Multi-agent MeanFlow Policy (VGM$^2$P), a simple yet effective flow-based policy learning framework that enables efficient action generation with coefficient-insensitive conditional behavior cloning. Specifically, VGM$^2$P uses global advantage values to guide agent collaboration, treating optimal policy learning as conditional behavior cloning. Additionally, to improve policy expressiveness and inference efficiency in multi-agent scenarios, it leverages classifier-free guidance MeanFlow for both policy training and execution. Experiments on tasks with both discrete and continuous action spaces demonstrate that, even when trained solely via conditional behavior cloning, VGM$^2$P efficiently achieves performance comparable to state-of-the-art methods. --- ### Exploration of Pareto-preserving Search Space Transformations in Multi-objective Test Functions Authors: Diederick Vermetten, Jeroen Rook Published: 2026-04-09 Categories: cs.NE arXiv: https://arxiv.org/abs/2604.08173v2 Benchmark problems are an important tool for gaining understanding of optimization algorithms. Since algorithms often aim to perform well on benchmarks, biases in benchmark design provide misleading insights. In single-objective optimization, for example, many problems used to have their optimum in the center of the search domain. To remedy these issues, search space transformations have been widely adopted by benchmark suites, preventing algorithms from exploiting unintended structure. In multi-objective optimization, problem design has focused primarily on the objective space structure. While this focus addresses important aspects of the multi-objective nature of the problems, the search space structures of these problems have received comparatively limited attention. In this work, we re-emphasize the importance of transformations in the search space, and address the challenges inherent in adding transformations to boundary constraints problems without impacting the structure of the objective space. We utilized two parameterized, bijective transformations to create different instantiations of popular benchmark problems, and show how these changes impact the performance of various multi-objective optimization algorithms. In addition to the search space transformations, we show that such parameterized transformations can also be applied to the objective space, and compare their respective performance impacts. --- ### On the Global Photometric Alignment for Low-Level Vision Authors: Mingjia Li, Tianle Du, Hainuo Wang, Qiming Hu, Xiaojie Guo Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08172v1 Supervised low-level vision models rely on pixel-wise losses against paired references, yet paired training sets exhibit per-pair photometric inconsistency, say, different image pairs demand different global brightness, color, or white-balance mappings. This inconsistency enters through task-intrinsic photometric transfer (e.g., low-light enhancement) or unintended acquisition shifts (e.g., de-raining), and in either case causes an optimization pathology. Standard reconstruction losses allocate disproportionate gradient budget to conflicting per-pair photometric targets, crowding out content restoration. In this paper, we investigate this issue and prove that, under least-squares decomposition, the photometric and structural components of the prediction-target residual are orthogonal, and that the spatially dense photometric component dominates the gradient energy. Motivated by this analysis, we propose Photometric Alignment Loss (PAL). This flexible supervision objective discounts nuisance photometric discrepancy via closed-form affine color alignment while preserving restoration-relevant supervision, requiring only covariance statistics and tiny matrix inversion with negligible overhead. Across 6 tasks, 16 datasets, and 16 architectures, PAL consistently improves metrics and generalization. The implementation is in the appendix. --- ### OceanMAE: A Foundation Model for Ocean Remote Sensing Authors: Viola-Joanna Stamer, Panagiotis Agrafiotis, Behnood Rasti, Begüm Demir Published: 2026-04-09 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2604.08171v1 Accurate ocean mapping is essential for applications such as bathymetry estimation, seabed characterization, marine litter detection, and ecosystem monitoring. However, ocean remote sensing (RS) remains constrained by limited labeled data and by the reduced transferability of models pre-trained mainly on land-dominated Earth observation imagery. In this paper, we propose OceanMAE, an ocean-specific masked autoencoder that extends standard MAE pre-training by integrating multispectral Sentinel-2 observations with physically meaningful ocean descriptors during self-supervised learning. By incorporating these auxiliary ocean features, OceanMAE is designed to learn more informative and ocean-aware latent representations from large- scale unlabeled data. To transfer these representations to downstream applications, we further employ a modified UNet-based framework for marine segmentation and bathymetry estimation. Pre-trained on the Hydro dataset, OceanMAE is evaluated on MADOS and MARIDA for marine pollutant and debris segmentation, and on MagicBathyNet for bathymetry regression. The experiments show that OceanMAE yields the strongest gains on marine segmentation, while bathymetry benefits are competitive and task-dependent. In addition, an ablation against a standard MAE on MARIDA indicates that incorporating auxiliary ocean descriptors during pre-training improves downstream segmentation quality. These findings highlight the value of physically informed and domain-aligned self-supervised pre- training for ocean RS. Code and weights are publicly available at https://git.tu-berlin.de/joanna.stamer/SSLORS2. --- ### Activation Steering for Aligned Open-ended Generation without Sacrificing Coherence Authors: Niklas Herbster, Martin Zborowski, Alberto Tosato, Gauthier Gidel, Tommaso Tosato Published: 2026-04-09 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.08169v3 Alignment in LLMs is more brittle than commonly assumed: misalignment can be induced by adversarial prompts, benign fine-tuning, emergent misalignment, and goal misgeneralization. Recent evidence suggests that some misalignment behaviors are encoded as linear structure in activation space, making it tractable via activation steering, which could be used as a lightweight runtime defense. We implement three methods: Steer-With-Fixed-Coefficient (SwFC), which applies uniform additive steering, and two novel projection-aware methods, Steer-to-Target-Projection (StTP) and Steer-to-Mirror-Projection (StMP), that use a logistic regression decision boundary to selectively intervene only on tokens whose activations fall below the threshold. We evaluate these methods on two threat models, dishonesty and dismissiveness, using malicious system prompts as a controlled proxy for misalignment. We conduct our experiments on two architectures (Llama-3.3-70B-Instruct and Qwen3.6-27B). All methods substantially recover alignment. StTP and StMP preserve general capabilities (MMLU, MT-Bench, AlpacaEval) better than uniform steering. Finally, we show that our honesty steering generalizes to out-of-distribution scenarios: a single honesty direction extracted from the aligned model significantly raises scores on the MASK benchmark, suppresses deception in multi-agent settings (Among Us), doubles the hidden-behavior discovery rate on AuditBench, and restores honesty in an emergently misaligned model. --- ### Activation Steering for Aligned Open-ended Generation without Sacrificing Coherence Authors: Niklas Herbster, Martin Zborowski, Alberto Tosato, Gauthier Gidel, Tommaso Tosato Published: 2026-04-09 Categories: cs.AI arXiv: https://arxiv.org/abs/2604.08169v2 Alignment in LLMs is more brittle than commonly assumed: misalignment can be induced by adversarial prompts, benign fine-tuning, emergent misalignment, and goal misgeneralization. Recent evidence suggests that some misalignment behaviors are encoded as linear structure in activation space, making it tractable via activation steering, which could be used as a lightweight runtime defense. We implement three methods: Steer-With-Fixed-Coefficient (SwFC), which applies uniform additive steering, and two novel projection-aware methods, Steer-to-Target-Projection (StTP) and Steer-to-Mirror-Projection (StMP), that use a logistic regression decision boundary to selectively intervene only on tokens whose activations fall below the threshold. We evaluate these methods on two threat models, dishonesty and dismissiveness, using malicious system prompts as a controlled proxy for misalignment. We conduct our experiments on two architectures (Llama-3.3-70B-Instruct and Qwen3.6-27B). All methods substantially recover alignment. StTP and StMP preserve general capabilities (MMLU, MT-Bench, AlpacaEval) better than uniform steering. Finally, we show that our honesty steering generalizes to out-of-distribution scenarios: a single honesty direction extracted from the aligned model significantly raises scores on the MASK benchmark, suppresses deception in multi-agent settings (Among Us), doubles the hidden-behavior discovery rate on AuditBench, and restores honesty in an emergently misaligned model --- ### ViVa: A Video-Generative Value Model for Robot Reinforcement Learning Authors: Jindi Lv, Hao Li, Jie Li, Fankun Kong, Yang Wang, Pengfei Yi, Yifei Nie, Xiaofeng Wang, Zheng Zhu, Chaojun Ni, Qiuping Deng, Hengtao Li et al. Published: 2026-04-09 Categories: cs.RO, cs.AI arXiv: https://arxiv.org/abs/2604.08168v2 Vision-language-action (VLA) models have advanced robot manipulation through large-scale pretraining, but real-world deployment remains challenging due to partial observability and delayed feedback. Reinforcement learning addresses this via value functions, which assess task progress and guide policy improvement. However, existing value models built on vision-language models (VLMs) struggle to capture temporal dynamics and physical interactions, undermining reliable value estimation in long-horizon tasks. In this paper, we propose ViVa, a video-generative value model that repurposes a pretrained video generator to jointly predict future proprioception and a scalar value. By grounding value estimation in anticipated embodiment dynamics, ViVa leverages spatiotemporal priors to intrinsically couple value with foresight beyond static snapshots. ViVa achieves state-of-the-art results in metric-based evaluation across three tasks, producing reliable value signals that accurately track task progress and detect execution errors. Integrated into RECAP, it achieves an average success rate of 80%, highlighting the promise of video-generative models for value estimation. --- ### T-Gated Adapter: A Lightweight Temporal Adapter for Vision-Language Medical Segmentation Authors: Pranjal Khadka Published: 2026-04-09 Categories: cs.CV arXiv: https://arxiv.org/abs/2604.08167v1 Medical image segmentation traditionally relies on fully supervised 3D architectures that demand a large amount of dense, voxel-level annotations from clinical experts which is a prohibitively expensive process. Vision Language Models (VLMs) offer a powerful alternative by leveraging broad visual semantic representations learned from billions of images. However, when applied independently to 2D slices of a 3D scan, these models often produce noisy and anatomically implausible segmentations that violate the inherent continuity of anatomical structures. We propose a temporal adapter that addresses this by injecting adjacent-slice context directly into the model's visual token representations. The adapter comprises a temporal transformer attending across a fixed context window at the token level, a spatial context block refining within-slice representations, and an adaptive gate balancing temporal and single-slice features. Training on 30 labeled volumes from the FLARE22 dataset, our method achieves a mean Dice of 0.704 across 13 abdominal organs with a gain of +0.206 over the baseline VLM trained with no temporal context. Zero-shot evaluation on BTCV and AMOS22 datasets yields consistent improvements of +0.210 and +0.230, with the average cross-domain performance drop reducing from 38.0% to 24.9%. Furthermore, in a cross-modality evaluation on AMOS22 MRI with neither model receiving any MRI supervision, our method achieves a mean Dice of 0.366, outperforming a fully supervised 3D baseline (DynUNet, 0.224) trained exclusively on CT, suggesting that CLIP's visual semantic representations generalize more gracefully across imaging modalities than convolutional features. --- ### Shift- and stretch-invariant non-negative matrix factorization with an application to brain tissue delineation in emission tomography data Authors: Anders S. Olsen, Miriam L. Navarro, Claus Svarer, Jesper L. Hinrich, Morten Mørup, Gitte M. Knudsen Published: 2026-04-09 Categories: cs.LG arXiv: https://arxiv.org/abs/2604.08161v1 Dynamic neuroimaging data, such as emission tomography measurements of radiotracer transport in blood or cerebrospinal fluid, often exhibit diffusion-like properties. These introduce distance-dependent temporal delays, scale-differences, and stretching effects that limit the effectiveness of conventional linear modeling and decomposition methods. To address this, we present the shift- and stretch-invariant non-negative matrix factorization framework. Our approach estimates both integer and non-integer temporal shifts as well as temporal stretching, all implemented in the frequency domain, where shifts correspond to phase modifications, and where stretching is handled via zero-padding or truncation. The model is implemented in PyTorch (https://github.com/anders-s-olsen/shiftstretchNMF). We demonstrate on synthetic data and brain emission tomography data that the model is able to account for stretching to provide more detailed characterization of brain tissue structure. ---