# Research Papers — 2026-03 · bundle page 68 of 243 > Full-text markdown of papers #3351–3400 of 12123 published in 2026-03, newest first. AI-readable mirrors of the full 12123-paper 2026-03 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2026-03/page/68 [← Previous bundle](https://aipod.dev/papers/archive/2026-03/page/67.md) [Next bundle →](https://aipod.dev/papers/archive/2026-03/page/69.md) ### Extending Precipitation Nowcasting Horizons via Spectral Fusion of Radar Observations and Foundation Model Priors Authors: Yuze Qin, Qingyong Li, Zhiqing Guo, Wen Wang, Yan Liu, Yangli-ao Geng Published: 2026-03-23 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2603.21768v3 Precipitation nowcasting is critical for disaster mitigation and aviation safety. However, radar-only models frequently suffer from a lack of large-scale atmospheric context, leading to performance degradation at longer lead times. While integrating meteorological variables predicted by weather foundation models offers a potential remedy, existing architectures fail to reconcile the profound representational heterogeneities between radar imagery and meteorological data. To bridge this gap, we propose PW-FouCast, a novel frequency-domain fusion framework that leverages Pangu-Weather forecasts as spectral priors within a Fourier-based backbone. Our architecture introduces three key innovations: (i) Pangu-Weather-guided Frequency Modulation to align spectral magnitudes and phases with meteorological priors; (ii) Frequency Memory to correct phase discrepancies and preserve temporal evolution; and (iii) Inverted Frequency Attention to reconstruct high-frequency details typically lost in spectral filtering. Extensive experiments on the SEVIR and MeteoNet benchmarks demonstrate that PW-FouCast achieves state-of-the-art performance, effectively extending the reliable forecast horizon while maintaining structural fidelity. Our code is available at https://github.com/Onemissed/PW-FouCast. --- ### Cycle Inverse-Consistent TransMorph: A Balanced Deep Learning Framework for Brain MRI Registration Authors: Jiaqi Shang, Haojin Wu, Yinyi Lai, Zongyu Li, Chenghao Zhang, Jia Guo Published: 2026-03-23 Categories: eess.IV, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2603.21760v1 Deformable image registration plays a fundamental role in medical image analysis by enabling spatial alignment of anatomical structures across subjects. While recent deep learning-based approaches have significantly improved computational efficiency, many existing methods remain limited in capturing long-range anatomical correspondence and maintaining deformation consistency. In this work, we present a cycle inverse-consistent transformer-based framework for deformable brain MRI registration. The model integrates a Swin-UNet architecture with bidirectional consistency constraints, enabling the joint estimation of forward and backward deformation fields. This design allows the framework to capture both local anatomical details and global spatial relationships while improving deformation stability. We conduct a comprehensive evaluation of the proposed framework on a large multi-center dataset consisting of 2851 T1-weighted brain MRI scans aggregated from 13 public datasets. Experimental results demonstrate that the proposed framework achieves strong and balanced performance across multiple quantitative evaluation metrics while maintaining stable and physically plausible deformation fields. Detailed quantitative comparisons with baseline methods, including ANTs, ICNet, and VoxelMorph, are provided in the appendix. Experimental results demonstrate that CICTM achieves consistently strong performance across multiple evaluation criteria while maintaining stable and physically plausible deformation fields. These properties make the proposed framework suitable for large-scale neuroimaging datasets where both accuracy and deformation stability are critical. --- ### Let's Think with Images Efficiently! An Interleaved-Modal Chain-of-Thought Reasoning Framework with Dynamic and Precise Visual Thoughts Authors: Xu Liu, Yongheng Zhang, Qiguang Chen, Yao Li, Sheng Wang, Libo Qin Published: 2026-03-23 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2603.21754v1 Recently, Interleaved-modal Chain-of-Thought (ICoT) reasoning has achieved remarkable success by leveraging both multimodal inputs and outputs, attracting increasing attention. While achieving promising performance, current ICoT methods still suffer from two major limitations: (1) Static Visual Thought Positioning, which statically inserts visual information at fixed steps, resulting in inefficient and inflexible reasoning; and (2) Broken Visual Thought Representation, which involves discontinuous and semantically incoherent visual tokens. To address these limitations, we introduce Interleaved-modal Chain-of-Thought reasoning with Dynamic and Precise Visual Thoughts (DaP-ICoT), which incorporates two key components: (1) Dynamic Visual Thought Integration adaptively introduces visual inputs based on reasoning needs, reducing redundancy and improving efficiency. (2) Precise Visual Thought Guidance ensures visual semantically coherent and contextually aligned representations. Experiments across multiple benchmarks and models demonstrate that DaP-ICoT achieves state-of-the-art performance. In addition, DaP-ICoT significantly reduces the number of inserted images, leading to a 72.6% decrease in token consumption, enabling more efficient ICoT reasoning. --- ### Identifiability and amortized inference limitations in Kuramoto models Authors: Emma Hannula, Jana de Wiljes, Matthew T. Moores, Heikki Haario, Lassi Roininen Published: 2026-03-23 Categories: stat.AP, cs.LG arXiv: https://arxiv.org/abs/2603.21752v1 Bayesian inference is a powerful tool for parameter estimation and uncertainty quantification in dynamical systems. However, for nonlinear oscillator networks such as Kuramoto models, widely used to study synchronization phenomena in physics, biology, and engineering, inference is often computationally prohibitive due to high-dimensional state spaces and intractable likelihood functions. We present an amortized Bayesian inference approach that learns a neural approximation of the posterior from simulated phase dynamics, enabling fast, scalable inference without repeated sampling or optimization. Applied to synthetic Kuramoto networks, the method shows promising results in approximating posterior distributions and capturing uncertainty, with computational savings compared to traditional Bayesian techniques. These findings suggest that amortized inference is a practical and flexible framework for uncertainty-aware analysis of oscillator networks. --- ### Model selection in hybrid quantum neural networks with applications to quantum transformer architectures Authors: Harsh Wadhwa, Rahul Bhowmick, Naipunnya Raj, Rajiv Sangle, Ruchira V. Bhat, Krishnakumar Sabapathy Published: 2026-03-23 Categories: quant-ph, cs.LG arXiv: https://arxiv.org/abs/2603.21749v1 Quantum machine learning models generally lack principled design guidelines, often requiring full resource-intensive training across numerous choices of encodings, quantum circuit designs and initialization strategies to find effective configuration. To address this challenge, we develope the Quantum Bias-Expressivity Toolbox ($\texttt{QBET}$), a framework for evaluating quantum, classical, and hybrid transformer architectures. In this toolbox, we introduce lean metrics for Simplicity Bias ($\texttt{SB}$) and Expressivity ($\texttt{EXP}$), for comparing across various models, and extend the analysis of $\texttt{SB}$ to generative and multiclass-classification tasks. We show that $\texttt{QBET}$ enables efficient pre-screening of promising model variants obviating the need to execute complete training pipelines. In evaluations on transformer-based classification and generative tasks we employ a total of $18$ qubits for embeddings ($6$ qubits each for query, key, and value). We identify scenarios in which quantum self-attention variants surpass their classical counterparts by ranking the respective models according to the $\texttt{SB}$ metric and comparing their relative performance. --- ### Getting to the Point: Pointing Improves LVLMs at Counting Authors: Simone Alghisi, Massimo Rizzoli, Seyed Mahed Mousavi, Giuseppe Riccardi Published: 2026-03-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.21746v2 Pointing-based methods decompose complex tasks as sequential grounding and reasoning steps. Given a query, the model first grounds the relevant objects by generating their coordinates, and then predicts an answer conditioned on these points. While this approach has been shown to increase the performance of Large Vision-Language Models (LVLMs), it remains unclear why and how it improves the models' visual reasoning. In this work, we evaluate pointing-based methods in the task of zero-shot counting in visual scenes. We experiment with multiple fine-tuning and training-free approaches on state-of-the-art LVLMs, and compare them with Point-then-Count (PtC), where models first generate point coordinates for the target objects and then predict their count. Our results show that PtC achieves the highest accuracy among the evaluated approaches, with predicted points correctly grounded in the image in more than 94% of cases (based on F1-score). Mechanistic analyses show that gains arise from spatial information encoded in the predicted coordinates. Nevertheless, grounding performance varies across image regions, revealing spatial biases. Finally, the results indicate that PtC improves out-of-distribution generalization on both synthetic and real data, suggesting the potential of coordinates to help LVLMs improve their counting skills. --- ### The Presupposition Problem in Representation Genesis Authors: Yiling Wu Published: 2026-03-23 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2603.21745v2 Large language models are the first systems to achieve high cognitive performance without clearly undergoing representation genesis: the transition from a non-representing physical system to one whose states guide behavior in a content-sensitive way. Prior cognitive systems had already made this transition before we could examine it, and philosophy of mind treated genesis as a background condition rather than an explanatory target. LLMs provide a case that does not clearly involve this transition, making the genesis question newly urgent: if genesis did not occur, which cognitive capacities are affected, and why? We currently lack the conceptual resources to answer this. The reason, this paper argues, is structural. Major frameworks in philosophy of mind, including the Language of Thought hypothesis, teleosemantics, predictive processing, enactivism, and genetic phenomenology, share a common feature when applied to the genesis question: at some explanatory step, each deploys concepts whose explanatory purchase depends on the system already being organized as a representer. This pattern, which we call the Representation Presupposition structure, generates systematic explanatory deferral. Attempts to explain the first acquisition of content-manipulable representation within the existing categorical vocabulary import resources from the representational side of the transition itself. We call this the Representation Regress. The paper offers a conceptual diagnosis rather than a new theory, establishing the structure of the problem and deriving two minimum adequacy conditions for any account that avoids this pattern. LLMs make the absence of such a theory consequential rather than merely theoretical. --- ### CellFluxRL: Biologically-Constrained Virtual Cell Modeling via Reinforcement Learning Authors: Dongxia Wu, Shiye Su, Yuhui Zhang, Elaine Sui, Emma Lundberg, Emily B. Fox, Serena Yeung-Levy Published: 2026-03-23 Categories: cs.LG, q-bio.QM arXiv: https://arxiv.org/abs/2603.21743v4 Building virtual cells with generative models to simulate cellular behavior in silico is emerging as a promising paradigm for accelerating drug discovery. However, prior image-based generative approaches can produce implausible cell images that violate basic physical and biological constraints. To address this, we propose to post-train virtual cell models with reinforcement learning (RL), leveraging biologically meaningful evaluators as reward functions. We design seven rewards spanning three categories-biological function, structural validity, and morphological correctness-and optimize the state-of-the-art CellFlux model to yield CellFluxRL. CellFluxRL consistently improves over CellFlux across all rewards, with further performance boosts from test-time scaling. Overall, our results present a virtual cell modeling framework that enforces physically-based constraints through RL, advancing beyond "visually realistic" generations towards "biologically meaningful" ones. --- ### The Reasoning Error About Reasoning: Why Different Types of Reasoning Require Different Representational Structures Authors: Yiling Wu Published: 2026-03-23 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2603.21736v2 Different types of reasoning impose different structural demands on representational systems, yet no systematic account of these demands exists across psychology, AI, and philosophy of mind. I propose a framework identifying four structural properties of representational systems: operability, consistency, structural preservation, and compositionality. These properties are demanded to different degrees by different forms of reasoning, from induction through analogy and causal inference to deduction and formal logic. Each property excludes a distinct class of reasoning failure. The analysis reveals a principal structural boundary: reasoning types below it can operate on associative, probabilistic representations, while those above it require all four properties to be fully satisfied. Scaling statistical learning without structural reorganization is insufficient to cross this boundary, because the structural guarantees required by deductive reasoning cannot be approximated through probabilistic means. Converging evidence from AI evaluation, developmental psychology, and cognitive neuroscience supports the framework at different levels of directness. Three testable predictions are derived, including compounding degradation, selective vulnerability to targeted structural disruption, and irreducibility under scaling. The framework is a necessary-condition account, agnostic about representational format, that aims to reorganize existing debates rather than close them. --- ### Cognitive Agency Surrender: Defending Epistemic Sovereignty via Scaffolded AI Friction Authors: Kuangzhe Xu, Yu Shen, Longjie Yan, Yinghui Ren Published: 2026-03-23 Categories: cs.HC, cs.AI arXiv: https://arxiv.org/abs/2603.21735v2 The proliferation of Generative Artificial Intelligence has transformed benign cognitive offloading into a systemic risk of cognitive agency surrender. Driven by the commercial dogma of "zero-friction" design, highly fluent AI interfaces actively exploit human cognitive miserliness, prematurely satisfying the need for cognitive closure and inducing severe automation bias. To empirically quantify this epistemic erosion, we deployed a zero-shot semantic classification pipeline ($τ=0.7$) on 1,223 high-confidence AI-HCI papers from 2023 to early 2026. Our analysis reveals an escalating "agentic takeover": a brief 2025 surge in research defending human epistemic sovereignty (19.1%) was abruptly suppressed in early 2026 (13.1%) by an explosive shift toward optimizing autonomous machine agents (19.6%), while frictionless usability maintained a structural hegemony (67.3%). To dismantle this trap, we theorize "Scaffolded Cognitive Friction," repurposing Multi-Agent Systems (MAS) as explicit cognitive forcing functions (e.g., computational Devil's Advocates) to inject germane epistemic tension and disrupt heuristic execution. Furthermore, we outline a multimodal computational phenotyping agenda -- integrating gaze transition entropy, task-evoked pupillometry, fNIRS, and Hierarchical Drift Diffusion Modeling (HDDM) -- to mathematically decouple decision outcomes from cognitive effort. Ultimately, intentionally designed friction is not merely a psychological intervention, but a foundational technical prerequisite for enforcing global AI governance and preserving societal cognitive resilience. --- ### EvoIdeator: Evolving Scientific Ideas through Checklist-Grounded Reinforcement Learning Authors: Andreas Sauter, Yuyue Zhao, Jacopo Urbani, Wenxiang Hu, Zaiqiao Meng, Lun Zhou, Xiaohui Yan, Yougang Lyu Published: 2026-03-23 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2603.21728v1 Scientific idea generation is a cornerstone of autonomous knowledge discovery, yet the iterative evolution required to transform initial concepts into high-quality research proposals remains a formidable challenge for Large Language Models (LLMs). Existing Reinforcement Learning (RL) paradigms often rely on rubric-based scalar rewards that provide global quality scores but lack actionable granularity. Conversely, language-based refinement methods are typically confined to inference-time prompting, targeting models that are not explicitly optimized to internalize such critiques. To bridge this gap, we propose \textbf{EvoIdeator}, a framework that facilitates the evolution of scientific ideas by aligning the RL training objective with \textbf{checklist-grounded feedback}. EvoIdeator leverages a structured judge model to generate two synergistic signals: (1) \emph{lexicographic rewards} for multi-dimensional optimization, and (2) \emph{fine-grained language feedback} that offers span-level critiques regarding grounding, feasibility, and methodological rigor. By integrating these signals into the RL loop, we condition the policy to systematically utilize precise feedback during both optimization and inference. Extensive experiments demonstrate that EvoIdeator, built on Qwen3-4B, significantly outperforms much larger frontier models across key scientific metrics. Crucially, the learned policy exhibits strong generalization to diverse external feedback sources without further fine-tuning, offering a scalable and rigorous path toward self-refining autonomous ideation. --- ### CurvZO: Adaptive Curvature-Guided Sparse Zeroth-Order Optimization for Efficient LLM Fine-Tuning Authors: Shuo Wang, Ziyu Chen, Ming Tang Published: 2026-03-23 Categories: cs.AI, cs.LG arXiv: https://arxiv.org/abs/2603.21725v1 Fine-tuning large language models (LLMs) with backpropagation achieves high performance but incurs substantial memory overhead, limiting scalability on resource-constrained hardware. Zeroth-order (ZO) optimization provides a memory-efficient alternative by relying solely on forward passes, yet it typically suffers from slow or unstable convergence due to high-variance gradient estimates. Sparse ZO updates partially address this issue by perturbing only a subset of parameters, but their effectiveness hinges on selecting informative parameters, which is challenging in ZO optimization because each query yields only scalar feedback. We propose \textbf{Adaptive Curvature-Guided Sparse Zeroth-Order Optimization (CurvZO)}, which tracks curvature signals online from scalar ZO feedback and leverages these signals to construct a parameter-wise sampling distribution for selecting coordinates at each update, reducing the variance of the sparse ZO gradient estimator. Moreover, CurvZO dynamically adapts the perturbation budget to the evolving curvature signal distribution, yielding sparse ZO updates that remain both focused and sufficiently exploratory. Extensive experiments on OPT and Llama across diverse NLP tasks show that CurvZO consistently improves fine-tuning performance and reduces training time over ZO baselines. It improves accuracy by up to 4.4 points and achieves up to a $2\times$ speedup, while preserving memory efficiency. --- ### FISformer: Replacing Self-Attention with a Fuzzy Inference System in Transformer Models for Time Series Forecasting Authors: Bulent Haznedar, Levent Karacan Published: 2026-03-23 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2603.21724v1 Transformers have achieved remarkable progress in time series forecasting, yet their reliance on deterministic dot-product attention limits their capacity to model uncertainty and nonlinear dependencies across multivariate temporal dimensions. To address this limitation, we propose FISFormer, a Fuzzy Inference System-driven Transformer that replaces conventional attention with a FIS Interaction mechanism. In this framework, each query-key pair undergoes a fuzzy inference process for every feature dimension, where learnable membership functions and rule-based reasoning estimate token-wise relational strengths. These FIS-derived interaction weights capture uncertainty and provide interpretable, continuous mappings between tokens. A softmax operation is applied along the token axis to normalize these weights, which are then combined with the corresponding value features through element-wise multiplication to yield the final context-enhanced token representations. This design fuses the interpretability and uncertainty modeling of fuzzy logic with the representational power of Transformers. Extensive experiments on multiple benchmark datasets demonstrate that FISFormer achieves superior forecasting accuracy, noise robustness, and interpretability compared to state-of-the-art Transformer variants, establishing fuzzy inference as an effective alternative to conventional attention mechanisms. --- ### SemEval-2026 Task 12: Abductive Event Reasoning: Towards Real-World Event Causal Inference for Large Language Models Authors: Pengfei Cao, Mingxuan Yang, Yubo Chen, Chenlong Zhang, Mingxuan Liu, Kang Liu, Jun Zhao Published: 2026-03-23 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2603.21720v1 Understanding why real-world events occur is important for both natural language processing and practical decision-making, yet direct-cause inference remains underexplored in evidence-rich settings. To address this gap, we organized SemEval-2026 Task 12: Abductive Event Reasoning (AER).\footnote{The task data is available at https://github.com/sooo66/semeval2026-task12-dataset.git} The task asks systems to identify the most plausible direct cause of a target event from supporting evidence. We formulate AER as an evidence-grounded multiple-choice benchmark that captures key challenges of real-world causal reasoning, including distributed evidence, indirect background factors, and semantically related but non-causal distractors. The shared task attracted 122 participants and received 518 submissions. This paper presents the task formulation, dataset construction pipeline, evaluation setup, and system results. AER provides a focused benchmark for abductive reasoning over real-world events and highlights challenges for future work on causal reasoning and multi-document understanding. --- ### Probing How Scalable Table Data Enhances General Long-Context Reasoning Authors: Huaibing Xie, Guoliang Zhao, Yang Liu, Shihan Dou, Siming Huang, Yanling Xiao, Shaolei Wang, Yiting Liu, Cheng Zhang, Shaofan Liu, Pluto Zhou Published: 2026-03-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.21719v1 As real-world tasks grow increasingly complex, long-context reasoning has become a core capability for Large Language Models (LLMs). However, few studies explore which data types are effective for long-context reasoning and why. We find that structured table data with periodic structures shows strong potential for long-context reasoning. Motivated by this observation, we mathematically analyze tabular dependency structures using mutual information, revealing periodic non-vanishing dependencies in table data. Furthermore, we systematically analyze the capabilities of structured table data, conduct relevant scaling experiments, and validate its underlying mechanisms for enhancing long-context reasoning, yielding several meaningful insights. Leveraging these insights, we propose a simple yet scalable pipeline(TableLong) for synthesizing high-quality, diverse, and verifiable structured table data to boost long-context reasoning via RL. Extensive experimental results demonstrate that table data significantly enhances the long-context reasoning capability of LLMs across multiple long-context benchmarks (+8.24\% on average), and even improves performance on out-of-domain benchmarks (+8.06\% on average). We hope that our insights provide practical guidance for effective post-training data to enhance long-context reasoning in LLMs. --- ### Uncertainty-Aware Distribution-to-Distribution Flow Matching for Scientific Imaging Authors: Dongxia Wu, Yuhui Zhang, Serena Yeung-Levy, Emma Lundberg, Emily B. Fox Published: 2026-03-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.21717v4 Distribution-to-distribution generative models support scientific imaging tasks ranging from modeling cellular perturbation responses to translating medical images across conditions. Trustworthy generation requires reliability, or generalization across labs, devices, and experimental conditions, and accountability, or detecting out-of-distribution cases where predictions may be unreliable. We leverage Stochastic Flow Matching (SFM), a marginal-preserving stochastic extension of flow matching for improved generalization under distribution shift. SFM augments deterministic flows with a diffusion term together with a learned score-based drift correction, retaining the learned transport marginals while modeling conditional variability. Building on this SFM framework, we introduce Bayesian Stochastic Flow Matching (BSFM) as a companion uncertainty quantification mechanism and develop AVUQ (Antithetic Variance-reduction Uncertainty Quantification) to approximately estimate epistemic and aleatoric uncertainty via sample-efficient antithetic sampling with approximate posterior inference. We further use AVUQ to yield anomaly scores for unreliable generation detection. Experiments on cellular imaging (BBBC021, JUMP) and brain fMRI (Theory of Mind) across diverse unseen scenarios show that SFM improves generalization while AVUQ provides effective uncertainty-based anomaly scores under practical sampling budgets. --- ### Mixture-Greedy for Online Generative Model Selection: Is UCB Necessary in Diversity-Aware Multi-Armed Bandits? Authors: Bahar Dibaei Nia, Farzan Farnia Published: 2026-03-23 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2603.21716v2 Efficient selection among multiple generative models is increasingly important in modern generative AI, where sampling from suboptimal models is costly. This problem can be viewed as a multi-armed bandit (MAB) task. Under diversity-aware evaluation scores, a non-degenerate mixture of generators can outperform any individual model, distinguishing this MAB setting from classical best-arm identification. Prior approaches incorporate an Upper Confidence Bound (UCB) exploration bonus into the mixture objective. However, across multiple datasets and evaluation metrics, we observe that the UCB term consistently slows convergence and reduces sample efficiency. In contrast, a simple Mixture-Greedy strategy without explicit UCB-type optimism converges faster and achieves even better performance, particularly for widely used metrics such as FID and Vendi where tight confidence bounds are difficult to construct. We provide theoretical insight explaining this behavior: under structural conditions, diversity-aware objectives induce implicit exploration by favoring interior mixtures, leading to sampling of all arms and sublinear regret guarantees for diversity-based objectives. These results suggest that in diversity-aware multi-armed bandits, e.g., for generative model selection, exploration can arise intrinsically from the objective's geometry. --- ### When Exploration Comes for Free with Mixture-Greedy: Do we need UCB in Diversity-Aware Multi-Armed Bandits? Authors: Bahar Dibaei Nia, Farzan Farnia Published: 2026-03-23 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2603.21716v1 Efficient selection among multiple generative models is increasingly important in modern generative AI, where sampling from suboptimal models is costly. This problem can be formulated as a multi-armed bandit task. Under diversity-aware evaluation metrics, a non-degenerate mixture of generators can outperform any individual model, distinguishing this setting from classical best-arm identification. Prior approaches therefore incorporate an Upper Confidence Bound (UCB) exploration bonus into the mixture objective. However, across multiple datasets and evaluation metrics, we observe that the UCB term consistently slows convergence and often reduces sample efficiency. In contrast, a simple \emph{Mixture-Greedy} strategy without explicit UCB-type optimism converges faster and achieves even better performance, particularly for widely used metrics such as FID and Vendi where tight confidence bounds are difficult to construct. We provide theoretical insight explaining this behavior: under transparent structural conditions, diversity-aware objectives induce implicit exploration by favoring interior mixtures, leading to linear sampling of all arms and sublinear regret guarantees for entropy-based, kernel-based, and FID-type objectives. These results suggest that in diversity-aware multi-armed bandits for generative model selection, exploration can arise intrinsically from the objective geometry, questioning the necessity of explicit confidence bonuses. --- ### Compensating Visual Insufficiency with Stratified Language Guidance for Long-Tail Class Incremental Learning Authors: Xi Wang, Xu Yang, Donghao Sun, Cheng Deng Published: 2026-03-23 Categories: cs.AI, cs.CV arXiv: https://arxiv.org/abs/2603.21708v1 Long-tail class incremental learning (LT CIL) remains highly challenging because the scarcity of samples in tail classes not only hampers their learning but also exacerbates catastrophic forgetting under continuously evolving and imbalanced data distributions. To tackle these issues, we exploit the informativeness and scalability of language knowledge. Specifically, we analyze the LT CIL data distribution to guide large language models (LLMs) in generating a stratified language tree that hierarchically organizes semantic information from coarse to fine grained granularity. Building upon this structure, we introduce stratified adaptive language guidance, which leverages learnable weights to merge multi-scale semantic representations, thereby enabling dynamic supervisory adjustment for tail classes and alleviating the impact of data imbalance. Furthermore, we introduce stratified alignment language guidance, which exploits the structural stability of the language tree to constrain optimization and reinforce semantic visual alignment, thereby alleviating catastrophic forgetting. Extensive experiments on multiple benchmarks demonstrate that our method achieves state of the art performance. --- ### Data-Free Layer-Adaptive Merging via Fisher Information for Long-to-Short Reasoning LLMs Authors: Tian Xia Published: 2026-03-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.21705v1 Model merging has emerged as a practical approach to combine capabilities of specialized large language models (LLMs) without additional training. In the Long-to-Short (L2S) scenario, merging a base model with a long-chain-of-thought reasoning model aims to preserve reasoning accuracy while reducing output length. Existing methods rely on Task Arithmetic and its variants, which implicitly assume that model outputs vary linearly with the merging coefficient -- an assumption we show is systematically violated in L2S settings. We provide the first theoretical justification for layer-adaptive merging: we prove that merging error is bounded by a term proportional to the per-layer Hessian norm (Proposition~1), and establish that the Fisher Information Matrix (FIM) is a principled, computable proxy for this bound via the Fisher-Hessian equivalence at local optima. Building on this theory, we propose \textbf{FIM-Merging}, which computes diagonal FIM using only random token inputs (no domain-specific calibration data required) and uses it to assign per-layer merging coefficients. On the 7B L2S benchmark, FIM-TIES achieves state-of-the-art performance on five out of six evaluation benchmarks, including a \textbf{+6.2} point gain on MATH500 over ACM-TIES (90.2 vs.\ 84.0), while requiring no calibration data. On the 1.5B benchmark, FIM-TIES achieves an average accuracy of \textbf{47.3}, surpassing the previous best ACM-TIES (43.3) by \textbf{+3.9} points, while reducing average response length by \textbf{91.9\%} relative to the long-CoT model. Our framework also provides a unified theoretical explanation for why existing layer-adaptive methods such as ACM empirically outperform uniform merging. --- ### Rethinking Token Reduction for Large Vision-Language Models Authors: Yi Wang, Haofei Zhang, Qihan Huang, Anda Cao, Gongfan Fang, Wei Wang, Xuan Jin, Jie Song, Mingli Song, Xinchao Wang Published: 2026-03-23 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2603.21701v1 Large Vision-Language Models (LVLMs) excel in visual understanding and reasoning, but the excessive visual tokens lead to high inference costs. Although recent token reduction methods mitigate this issue, they mainly target single-turn Visual Question Answering (VQA), leaving the more practical multi-turn VQA (MT-VQA) scenario largely unexplored. MT-VQA introduces additional challenges, as subsequent questions are unknown beforehand and may refer to arbitrary image regions, making existing reduction strategies ineffective. Specifically, current approaches fall into two categories: prompt-dependent methods, which bias toward the initial text prompt and discard information useful for subsequent turns; prompt-agnostic ones, which, though technically applicable to multi-turn settings, rely on heuristic reduction metrics such as attention scores, leading to suboptimal performance. In this paper, we propose a learning-based prompt-agnostic method, termed MetaCompress, overcoming the limitations of heuristic designs. We begin by formulating token reduction as a learnable compression mapping, unifying existing formats such as pruning and merging into a single learning objective. Upon this formulation, we introduce a data-efficient training paradigm capable of learning optimal compression mappings with limited computational costs. Extensive experiments on MT-VQA benchmarks and across multiple LVLM architectures demonstrate that MetaCompress achieves superior efficiency-accuracy trade-offs while maintaining strong generalization across dialogue turns. Our code is available at https://github.com/MArSha1147/MetaCompress. --- ### PPGL-Swarm: Integrated Multimodal Risk Stratification and Hereditary Syndrome Detection in Pheochromocytoma and Paraganglioma Authors: Zelin Liu, Xiangfu Yu, Jie Huang, Ge Wang, Yizhe Yuan, Zhenyu Yi, Jing Xie, Haotian Jiang, Lichi Zhang Published: 2026-03-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.21700v1 Pheochromocytomas and paragangliomas (PPGLs) are rare neuroendocrine tumors, of which 15-25% develop metastatic disease with 5-year survival rates reported as low as 34%. PPGL may indicate hereditary syndromes requiring stricter, syndrome-specific treatment and surveillance, but clinicians often fail to recognize these associations in routine care. Clinical practice uses GAPP score for PPGL grading, but several limitations remain for PPGL diagnosis: (1) GAPP scoring demands a high workload for clinician because it requires the manual evaluation of six independent components; (2) key components such as cellularity and Ki-67 are often evaluated with subjective criteria; (3) several clinically relevant metastatic risk factors are not captured by GAPP, such as SDHB mutations, which have been associated with reported metastatic rates of 35-75%. Agent-driven diagnostic systems appear promising, but most lack traceable reasoning for decision-making and do not incorporate domain-specific knowledge such as PPGL genotype information. To address these limitations, we present PPGL-Swarm, an agentic PPGL diagnostic system that generates a comprehensive report, including automated GAPP scoring (with quantified cellularity and Ki-67), genotype risk alerts, and multimodal report with integrated evidence. The system provides an auditable reasoning trail by decomposing diagnosis into micro-tasks, each assigned to a specialized agent. The gene and table agents use knowledge enhancement to better interpret genotype and laboratory findings, and during training we use reinforcement learning to refine tool selection and task assignment. --- ### A Job I Like or a Job I Can Get: Designing Job Recommender Systems Using Field Experiments Authors: Guillaume Bied, Philippe Caillou, Bruno Crépon, Christophe Gaillac, Elia Pérennes, Michèle Sebag Published: 2026-03-23 Categories: econ.EM, stat.ML arXiv: https://arxiv.org/abs/2603.21699v1 Recommendation systems (RSs) are increasingly used to guide job seekers on online platforms, yet the algorithms currently deployed are typically optimized for predictive objectives such as clicks, applications, or hires, rather than job seekers' welfare. We develop a job-search model with an application stage in which the value of a vacancy depends on two dimensions: the utility it delivers to the worker and the probability that an application succeeds. The model implies that welfare-optimal RSs rank vacancies by an expected-surplus index combining both, and shows why rankings based solely on utility, hiring probabilities, or observed application behavior are generically suboptimal, an instance of the inversion problem between behavior and welfare. We test these predictions and quantify their practical importance through two randomized field experiments conducted with the French public employment service. The first experiment, comparing existing algorithms and their combinations, provides behavioral evidence that both dimensions shape application decisions. Guided by the model and these results, the second experiment extends the comparison to an RS designed to approximate the welfare-optimal ranking. The experiments generate exogenous variation in the vacancies shown to job seekers, allowing us to estimate the model, validate its behavioral predictions, and construct a welfare metric. Algorithms informed by the model-implied optimal ranking substantially outperform existing approaches and perform close to the welfare-optimal benchmark. Our results show that embedding predictive tools within a simple job-search framework and combining it with experimental evidence yields recommendation rules with substantial welfare gains in practice. --- ### A Blueprint for Self-Evolving Coding Agents in Vehicle Aerodynamic Drag Prediction Authors: Jinhui Ren, Huaiming Li, Yabin Liu, Tao Li, Zhaokun Liu, Yujia Liang, Zengle Ge, Chufan Wu, Xiaomin Yuan, Danyu Liu, Annan Li, Jianmin Wu Published: 2026-03-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.21698v1 High-fidelity vehicle drag evaluation is constrained less by solver runtime than by workflow friction: geometry cleanup, meshing retries, queue contention, and reproducibility failures across teams. We present a contract-centric blueprint for self-evolving coding agents that discover executable surrogate pipelines for predicting drag coefficient $C_d$ under industrial constraints. The method formulates surrogate discovery as constrained optimization over programs, not static model instances, and combines Famou-Agent-style evaluator feedback with population-based island evolution, structured mutations (data, model, loss, and split policies), and multi-objective selection balancing ranking quality, stability, and cost. A hard evaluation contract enforces leakage prevention, deterministic replay, multi-seed robustness, and resource budgets before any candidate is admitted. Across eight anonymized evolutionary operators, the best system reaches a Combined Score of 0.9335 with sign-accuracy 0.9180, while trajectory and ablation analyses show that adaptive sampling and island migration are primary drivers of convergence quality. The deployment model is explicitly ``screen-and-escalate'': surrogates provide high-throughput ranking for design exploration, but low-confidence or out-of-distribution cases are automatically escalated to high-fidelity CFD. The resulting contribution is an auditable, reusable workflow for accelerating aerodynamic design iteration while preserving decision-grade reliability, governance traceability, and safety boundaries. --- ### Structured Visual Narratives Undermine Safety Alignment in Multimodal Large Language Models Authors: Rui Yang Tan, Yujia Hu, Roy Ka-Wei Lee Published: 2026-03-23 Categories: cs.CR, cs.AI, cs.MM arXiv: https://arxiv.org/abs/2603.21697v2 Multimodal Large Language Models (MLLMs) extend text-only LLMs with visual reasoning, but also introduce new safety failure modes under visually grounded instructions. We study comic-template jailbreaks that embed harmful goals inside simple three-panel visual narratives and prompt the model to role-play and "complete the comic." Building on JailbreakBench and JailbreakV, we introduce ComicJailbreak, a comic-based jailbreak benchmark with 1,167 attack instances spanning 10 harm categories and 5 task setups. Across 15 state-of-the-art MLLMs (six commercial and nine open-source), comic-based attacks achieve success rates comparable to strong rule-based jailbreaks and substantially outperform plain-text and random-image baselines, with ensemble success rates exceeding 90% on several commercial models. Then, with the existing defense methodologies, we show that these methods are effective against the harmful comics, they will induce a high refusal rate when prompted with benign prompts. Finally, using automatic judging and targeted human evaluation, we show that current safety evaluators can be unreliable on sensitive but non-harmful content. Our findings highlight the need for safety alignment robust to narrative-driven multimodal jailbreaks. --- ### MIND: Multi-agent inference for negotiation dialogue in travel planning Authors: Hunmin Do, Taejun Yoon, Kiyong Jung Published: 2026-03-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.21696v1 While Multi-Agent Debate (MAD) research has advanced, its efficacy in coordinating complex stakeholder interests such as travel planning remains largely unexplored. To bridge this gap, we propose MIND (Multi-agent Inference for Negotiation Dialogue), a framework designed to simulate realistic consensus-building among travelers with heterogeneous preferences. Grounded in the Theory of Mind (ToM), MIND introduces a Strategic Appraisal phase that infers opponent willingness (w) from linguistic nuances with 90.2% accuracy. Experimental results demonstrate that MIND outperforms traditional MAD frameworks, achieving a 20.5% improvement in High-w Hit and a 30.7% increase in Debate Hit-Rate, effectively prioritizing high-stakes constraints. Furthermore, qualitative evaluations via LLM-as-a-Judge confirm that MIND surpasses baselines in Rationality (68.8%) and Fluency (72.4%), securing an overall win rate of 68.3%. These findings validate that MIND effectively models human negotiation dynamics to derive persuasive consensus. --- ### RefracGS: Novel View Synthesis Through Refractive Water Surfaces with 3D Gaussian Ray Tracing Authors: Yiming Shao, Qiyu Dai, Chong Gao, Guanbin Li, Yequan Wang, He Sun, Qiong Zeng, Baoquan Chen, Wenzheng Chen Published: 2026-03-23 Categories: cs.CV, cs.GR arXiv: https://arxiv.org/abs/2603.21695v2 Novel view synthesis (NVS) through non-planar refractive surfaces presents fundamental challenges due to severe, spatially varying optical distortions. While recent representations like NeRF and 3D Gaussian Splatting (3DGS) excel at NVS, their assumption of straight-line ray propagation fails under these conditions, leading to significant artifacts. To overcome this limitation, we introduce RefracGS, a framework that jointly reconstructs the refractive water surface and the scene beneath the interface. Our key insight is to explicitly decouple the refractive boundary from the target objects: the refractive surface is modeled via a neural height field, capturing wave geometry, while the underlying scene is represented as a 3D Gaussian field. We formulate a refraction-aware Gaussian ray tracing approach that accurately computes non-linear ray trajectories using Snell's law and efficiently renders the underlying Gaussian field while backpropagating the loss gradients to the parameterized refractive surface. Through end-to-end joint optimization of both representations, our method ensures high-fidelity NVS and view-consistent surface recovery. Experiments on both synthetic and real-world scenes with complex waves demonstrate that RefracGS outperforms prior refractive methods in visual quality, while achieving 15x faster training and real-time rendering at 200 FPS. The project page for RefracGS is available at https://yimgshao.github.io/refracgs/. --- ### RefracGS: Novel View Synthesis Through Refractive Water Surfaces with 3D Gaussian Ray Tracing Authors: Yiming Shao, Qiyu Dai, Chong Gao, Guanbin Li, Yeqiang Wang, He Sun, Qiong Zeng, Baoquan Chen, Wenzheng Chen Published: 2026-03-23 Categories: cs.CV, cs.GR arXiv: https://arxiv.org/abs/2603.21695v1 Novel view synthesis (NVS) through non-planar refractive surfaces presents fundamental challenges due to severe, spatially varying optical distortions. While recent representations like NeRF and 3D Gaussian Splatting (3DGS) excel at NVS, their assumption of straight-line ray propagation fails under these conditions, leading to significant artifacts. To overcome this limitation, we introduce RefracGS, a framework that jointly reconstructs the refractive water surface and the scene beneath the interface. Our key insight is to explicitly decouple the refractive boundary from the target objects: the refractive surface is modeled via a neural height field, capturing wave geometry, while the underlying scene is represented as a 3D Gaussian field. We formulate a refraction-aware Gaussian ray tracing approach that accurately computes non-linear ray trajectories using Snell's law and efficiently renders the underlying Gaussian field while backpropagating the loss gradients to the parameterized refractive surface. Through end-to-end joint optimization of both representations, our method ensures high-fidelity NVS and view-consistent surface recovery. Experiments on both synthetic and real-world scenes with complex waves demonstrate that RefracGS outperforms prior refractive methods in visual quality, while achieving 15x faster training and real-time rendering at 200 FPS. The project page for RefracGS is available at https://yimgshao.github.io/refracgs/. --- ### Deterministic Hallucination Detection in Medical VQA via Confidence-Evidence Bayesian Gain Authors: Mohammad Asadi, Tahoura Nedaee, Jack W. O'Sullivan, Euan Ashley, Ehsan Adeli Published: 2026-03-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.21693v2 Multimodal large language models (MLLMs) have shown strong potential for medical Visual Question Answering (VQA), yet they remain prone to hallucinations, defined as generating responses that contradict the input image, posing serious risks in clinical settings. Current hallucination detection methods, such as Semantic Entropy (SE) and Vision-Amplified Semantic Entropy (VASE), require 10 to 20 stochastic generations per sample together with an external natural language inference model for semantic clustering, making them computationally expensive and difficult to deploy in practice. We observe that hallucinated responses exhibit a distinctive signature directly in the model's own log-probabilities: inconsistent token-level confidence and weak sensitivity to visual evidence. Based on this observation, we propose Confidence-Evidence Bayesian Gain (CEBaG), a deterministic hallucination detection method that requires no stochastic sampling, no external models, and no task-specific hyperparameters. CEBaG combines two complementary signals: token-level predictive variance, which captures inconsistent confidence across response tokens, and evidence magnitude, which measures how much the image shifts per-token predictions relative to text-only inference. Evaluated across four medical MLLMs and three VQA benchmarks (16 experimental settings), CEBaG achieves the highest AUC in 13 of 16 settings and improves over VASE by 8 AUC points on average, while being fully deterministic and self-contained. https://github.com/masadi-99/CEBaG --- ### Reasoning Provenance for Autonomous AI Agents: Structured Behavioral Analytics Beyond State Checkpoints and Execution Traces Authors: Neelmani Vispute, Aditya Kadam Published: 2026-03-23 Categories: cs.AI, cs.DC, cs.SE arXiv: https://arxiv.org/abs/2603.21692v2 As AI agents transition from human-supervised copilots to autonomous platform infrastructure, the ability to analyze their reasoning behavior across populations of investigations becomes a pressing infrastructure requirement. Existing operational tooling addresses adjacent needs effectively: state checkpoint systems enable fault tolerance; observability platforms provide execution traces for debugging; telemetry standards ensure interoperability. What current systems do not natively provide as a first-class, schema-level primitive is structured reasoning provenance -- normalized, queryable records of why the agent chose each action, what it concluded from each observation, how each conclusion shaped its strategy, and which evidence supports its final verdict. This paper introduces the Agent Execution Record (AER), a structured reasoning provenance primitive that captures intent, observation, and inference as first-class queryable fields on every step, alongside versioned plans with revision rationale, evidence chains, structured verdicts with confidence scores, and delegation authority chains. We formalize the distinction between computational state persistence and reasoning provenance, argue that the latter cannot in general be faithfully reconstructed from the former, and show how AERs enable population-level behavioral analytics: reasoning pattern mining, confidence calibration, cross-agent comparison, and counterfactual regression testing via mock replay. We present a domain-agnostic model with extensible domain profiles, a reference implementation and SDK, and outline an evaluation methodology informed by preliminary deployment on a production platformized root cause analysis agent. --- ### AI Token Futures Market: Commoditization of Compute and Derivatives Contract Design Authors: Yicai Xing Published: 2026-03-23 Categories: cs.AI, econ.GN arXiv: https://arxiv.org/abs/2603.21690v1 As large language models (LLMs) and vision-language-action models (VLAs) become widely deployed, the tokens consumed by AI inference are evolving into a new type of commodity. This paper systematically analyzes the commodity attributes of tokens, arguing for their transition from intelligent service outputs to compute infrastructure raw materials, and draws comparisons with established commodities such as electricity, carbon emission allowances, and bandwidth. Building on the historical experience of electricity futures markets and the theory of commodity financialization, we propose a complete design for standardized token futures contracts, including the definition of a Standard Inference Token (SIT), contract specifications, settlement mechanisms, margin systems, and market-maker regimes. By constructing a mean-reverting jump-diffusion stochastic process model and conducting Monte Carlo simulations, we evaluate the hedging efficiency of the proposed futures contracts for application-layer enterprises. Simulation results show that, under an application-layer demand explosion scenario, token futures can reduce enterprise compute cost volatility by 62%-78%. We also explore the feasibility of GPU compute futures and discuss the regulatory framework for token futures markets, providing a theoretical foundation and practical roadmap for the financialization of compute resources. --- ### MIRAGE: The Illusion of Visual Understanding Authors: Mohammad Asadi, Jack W. O'Sullivan, Fang Cao, Tahoura Nedaee, Kamyar Rajabalifardi, Fei-Fei Li, Ehsan Adeli, Euan Ashley Published: 2026-03-23 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.21687v3 Multimodal AI systems have achieved remarkable performance across a broad range of real-world tasks, yet the mechanisms underlying visual-language reasoning remain surprisingly poorly understood. We report three findings that challenge prevailing assumptions about how these systems process and integrate visual information. First, Frontier models readily generate detailed image descriptions and elaborate reasoning traces, including pathology-biased clinical findings, for images never provided; we term this phenomenon mirage reasoning. Second, without any image input, models also attain strikingly high scores across general and medical multimodal benchmarks, bringing into question their utility and design. In the most extreme case, our model achieved the top rank on a standard chest X-ray question-answering benchmark without access to any images. Third, when models were explicitly instructed to guess answers without image access, rather than being implicitly prompted to assume images were present, performance declined markedly. Explicit guessing appears to engage a more conservative response regime, in contrast to the mirage regime in which models behave as though images have been provided. These findings expose fundamental vulnerabilities in how visual-language models reason and are evaluated, pointing to an urgent need for private benchmarks that eliminate textual cues enabling non-visual inference, particularly in medical contexts where miscalibrated AI carries the greatest consequence. We introduce B-Clean as a principled solution for fair, vision-grounded evaluation of multimodal AI systems. --- ### LipsAM: Lipschitz-Continuous Amplitude Modifier for Audio Signal Processing and its Application to Plug-and-Play Dereverberation Authors: Kazuki Matsumoto, Ren Uchida, Kohei Yatabe Published: 2026-03-23 Categories: cs.SD, cs.LG arXiv: https://arxiv.org/abs/2603.21684v1 The robustness of deep neural networks (DNNs) can be certified through their Lipschitz continuity, which has made the construction of Lipschitz-continuous DNNs an active research field. However, DNNs for audio processing have not been a major focus due to their poor compatibility with existing results. In this paper, we consider the amplitude modifier (AM), a popular architecture for handling audio signals, and propose its Lipschitz-continuous variants, which we refer to as LipsAM. We prove a sufficient condition for an AM to be Lipschitz continuous and propose two architectures as examples of LipsAM. The proposed architectures were applied to a Plug-and-Play algorithm for speech dereverberation, and their improved stability is demonstrated through numerical experiments. --- ### Learning operators on labelled conditional distributions with applications to mean field control of non exchangeable systems Authors: Samy Mekkaoui, Huyên Pham, Xavier Warin Published: 2026-03-23 Categories: math.OC, math.PR, stat.ML arXiv: https://arxiv.org/abs/2603.21683v1 We study the approximation of operators acting on probability measures on a product space with prescribed marginal. Let $I$ be a label space endowed with a reference measure $λ$, and define $\cal M_λ$ as the set of probability measures on $I\times \mathbb{R}^d$ with first marginal $λ$. By disintegration, elements of $\cal M_λ$ correspond to families of labeled conditional distributions. Operators defined on this constrained measure space arise naturally in mean-field control problems with heterogeneous, non-exchangeable agents. Our main theoretical result establishes a universal approximation theorem for continuous operators on $\cal M_λ$. The proof combines cylindrical approximations of probability measures with DeepONet-type branch-trunk neural architecture, yielding finite-dimensional representations of such operators. We further introduce a sampling strategy for generating training measures in $\cal M_λ$, enabling practical learning of such conditional mean-field operators. We apply the method to the numerical resolution of mean-field control problems with heterogeneous interactions, thereby extending previous neural approaches developed for homogeneous (exchangeable) systems. Numerical experiments illustrate the accuracy and computational effectiveness of the proposed framework. --- ### CoNBONet: Conformalized Neuroscience-inspired Bayesian Operator Network for Reliability Analysis Authors: Shailesh Garg, Souvik Chakraborty Published: 2026-03-23 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2603.21678v1 Time-dependent reliability analysis of nonlinear dynamical systems under stochastic excitations is a critical yet computationally demanding task. Conventional approaches, such as Monte Carlo simulation, necessitate repeated evaluations of computationally expensive numerical solvers, leading to significant computational bottlenecks. To address this challenge, we propose \textit{CoNBONet}, a neuroscience-inspired surrogate model that enables fast, energy-efficient, and uncertainty-aware reliability analysis, providing a scalable alternative to techniques such as Monte Carlo simulations. CoNBONet, short for \textbf{Co}nformalized \textbf{N}euroscience-inspired \textbf{B}ayesian \textbf{O}perator \textbf{Net}work, leverages the expressive power of deep operator networks while integrating neuroscience-inspired neuron models to achieve fast, low-power inference. Unlike traditional surrogates such as Gaussian processes, polynomial chaos expansions, or support vector regression, that may face scalability challenges for high-dimensional, time-dependent reliability problems, CoNBONet offers \textit{fast and energy-efficient inference} enabled by a neuroscience-inspired network architecture, \textit{calibrated uncertainty quantification with theoretical guarantees} via split conformal prediction, and \textit{strong generalization capability} through an operator-learning paradigm that maps input functions to system response trajectories. Validation of the proposed CoNBONet for various nonlinear dynamical systems demonstrates that CoNBONet preserves predictive fidelity, and achieves reliable coverage of failure probabilities, making it a powerful tool for robust and scalable reliability analysis in engineering design. --- ### Thinking Deeper, Not Longer: Memory-Efficient Test-Time Reasoning with Depth-Recurrent Transformers for Compositional Generalization Authors: Hung-Hsuan Chen Published: 2026-03-23 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2603.21676v2 Standard Transformers have a fixed computational depth, limiting their ability to generalize to tasks that require variable-depth reasoning. The usual remedy, Chain-of-Thought (CoT), spends tokens to reason, inflating the key--value cache and making latency grow with the step count, so memory becomes the limiting cost when reasoning is served over large query batches. We study a depth-recurrent Transformer that decouples computational depth from parameter count by iterating a shared-weight block, so that each added reasoning step costs flat memory and linear latency, with no token generation. Three ingredients keep the recurrence stable for 20+ thinking steps: a silent thinking objective that supervises only the final output, LayerScale initialization, and an identity-biased gate that opens a gradient highway across steps. We characterize it on three compositional domains with decreasing structural bias: graph reachability (adjacency masking), nested boolean logic (relative positioning), and unstructured relational text (no positional cue). We find a \emph{computational frontier}: accuracy climbs once the thinking-step count meets the task's complexity, reaching near-perfect performance on the two structured tasks and a lower plateau on unstructured text. How it climbs depends on the structural bias---abruptly from chance on the graph task, gradually on the other two. Depth recurrence extrapolates beyond the training range: it succeeds on the graph task where fixed-depth models barely extrapolate, and on the two sequence tasks comes within two points of fixed-depth Transformers that use $4$--$6.4\times$ more parameters. On the graph task, whose adjacency mask makes propagation depth verifiable, intermediate per-step supervision---a standard recipe for deep iterative models---consistently \emph{harms} this extrapolation. We release the code for reproducibility. --- ### Thinking Deeper, Not Longer: Depth-Recurrent Transformers for Compositional Generalization Authors: Hung-Hsuan Chen Published: 2026-03-23 Categories: cs.LG, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2603.21676v1 Standard Transformers have a fixed computational depth, fundamentally limiting their ability to generalize to tasks requiring variable-depth reasoning, such as multi-hop graph traversal or nested logic. We propose a depth-recurrent Transformer that decouples computational depth from parameter count by iteratively applying a shared-weight Transformer block in latent space -- enabling the model to trade recurrence steps for deeper reasoning at inference time. Our architecture incorporates three mechanisms to make deep recurrence (20+ steps) stable: (1) a silent thinking objective that supervises only the final output, forcing genuine multi-step reasoning rather than intermediate heuristic shortcuts; (2) LayerScale initialization to protect fragile reasoning states from untrained layer noise; and (3) an identity-biased recurrence that creates a gradient highway across many steps. We evaluate on three compositional reasoning domains with decreasing inductive biases: graph reachability (strict adjacency masking), nested boolean logic (relative positioning), and unstructured relational text (where sequence position provides no structural hints). Across all tasks, we observe a clear \emph{computational frontier} -- a boundary where performance transitions from chance to near-perfect as thinking steps scale with task complexity. Moreover, these tasks reveal qualitatively different generalization behaviors: precise but brittle (graph), approximate but robust (logic), and autonomous latent routing without structural hints (text). This progression illuminates how the interplay between a task-invariant recurrent reasoning core and task-specific perceptual interfaces shapes out-of-distribution (OOD) generalization, offering a mechanistic perspective on vertical chain-of-thought that complements the prevailing horizontal token-generation paradigm. --- ### SPINONet: Scalable Spiking Physics-informed Neural Operator for Computational Mechanics Applications Authors: Shailesh Garg, Luis Mandl, Somdatta Goswami, Souvik Chakraborty Published: 2026-03-23 Categories: physics.comp-ph, cs.LG arXiv: https://arxiv.org/abs/2603.21674v1 Energy efficiency remains a critical challenge in deploying physics-informed operator learning models for computational mechanics and scientific computing, particularly in power-constrained settings such as edge and embedded devices, where repeated operator evaluations in dense networks incur substantial computational and energy costs. To address this challenge, we introduce the Separable Physics-informed Neuroscience-inspired Operator Network (SPINONet), a neuroscience-inspired framework that reduces redundant computation across repeated evaluations while remaining compatible with physics-informed training. SPINONet incorporates regression-friendly neuroscience-inspired spiking neurons through an architecture-aware design that enables sparse, event-driven computation, improving energy efficiency while preserving the continuous, coordinate-differentiable pathways required for computing spatio-temporal derivatives. We evaluate SPINONet on a range of partial differential equations representative of computational mechanics problems, with spatial, temporal, and parametric dependencies in both time-dependent and steady-state settings, and demonstrate predictive performance comparable to conventional physics-informed operator learning approaches despite the induced sparse communication. In addition, limited data supervision in a hybrid setup is shown to improve performance in challenging regimes where purely physics-informed training may converge to spurious solutions. Finally, we provide an analytical discussion linking architectural components and design choices of SPINONet to reductions in computational load and energy consumption. --- ### Optimizing Multi-Agent Weather Captioning via Text Gradient Descent: A Training-Free Approach with Consensus-Aware Gradient Fusion Authors: Shixu Liu Published: 2026-03-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.21673v1 Generating interpretable natural language captions from weather time series data remains a significant challenge at the intersection of meteorological science and natural language processing. While recent advances in Large Language Models (LLMs) have demonstrated remarkable capabilities in time series forecasting and analysis, existing approaches either produce numerical predictions without human-accessible explanations or generate generic descriptions lacking domain-specific depth. We introduce WeatherTGD, a training-free multi-agent framework that reinterprets collaborative caption refinement through the lens of Text Gradient Descent (TGD). Our system deploys three specialized LLM agents including a Statistical Analyst, a Physics Interpreter, and a Meteorology Expert that generate domain-specific textual gradients from weather time series observations. These gradients are aggregated through a novel Consensus-Aware Gradient Fusion mechanism that extracts common signals while preserving unique domain perspectives. The fused gradients then guide an iterative refinement process analogous to gradient descent, where each LLM-generated feedback signal updates the caption toward an optimal solution. Experiments on real-world meteorological datasets demonstrate that WeatherTGD achieves significant improvements in both LLM-based evaluation and human expert evaluation, substantially outperforming existing multi-agent baselines while maintaining computational efficiency through parallel agent execution. --- ### PRM-as-a-Judge: A Dense Evaluation Paradigm for Fine-Grained Robotic Auditing Authors: Yuheng Ji, Yuyang Liu, Huajie Tan, Xuchuan Huang, Fanding Huang, Yijie Xu, Cheng Chi, Yuting Zhao, Huaihai Lyu, Peterson Co, Mingyu Cao, Qiongyu Zhang et al. Published: 2026-03-23 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2603.21669v1 Current robotic evaluation is still largely dominated by binary success rates, which collapse rich execution processes into a single outcome and obscure critical qualities such as progress, efficiency, and stability. To address this limitation, we propose PRM-as-a-Judge, a dense evaluation paradigm that leverages Process Reward Models (PRMs) to audit policy execution directly from trajectory videos by estimating task progress from observation sequences. Central to this paradigm is the OPD (Outcome-Process-Diagnosis) metric system, which explicitly formalizes execution quality via a task-aligned progress potential. We characterize dense robotic evaluation through two axiomatic properties: macro-consistency, which requires additive and path-consistent aggregation, and micro-resolution, which requires sensitivity to fine-grained physical evolution. Under this formulation, potential-based PRM judges provide a natural instantiation of dense evaluation, with macro-consistency following directly from the induced scalar potential. We empirically validate the micro-resolution property using RoboPulse, a diagnostic benchmark specifically designed for probing micro-scale progress discrimination, where several trajectory-trained PRM judges outperform discriminative similarity-based methods and general-purpose foundation-model judges. Finally, leveraging PRM-as-a-Judge and the OPD metric system, we conduct a structured audit of mainstream policy paradigms across long-horizon tasks, revealing behavioral signatures and failure modes that are invisible to outcome-only metrics. --- ### Optimal Memory Encoding Through Fluctuation-Response Structure Authors: Lianxiang Cui, Kohei Nakajima, Kazuyuki Aihara Published: 2026-03-23 Categories: cs.NE arXiv: https://arxiv.org/abs/2603.21666v1 Physical reservoir computing exploits the intrinsic dynamics of physical systems for information processing, while keeping the internal dynamics fixed and training only linear readouts; yet the role of input encoding remains poorly understood. We show that optimal input encoding is a geometric problem governed by the system's fluctuation-response structure. By measuring steady-state fluctuations and linear response, we derive an analytical criterion for the input direction that maximizes task-specific linear memory under a fixed power constraint, termed Response-based Optimal Memory Encoding (ROME). Backpropagation-based encoder optimization is shown to be equivalent to ROME, revealing a trade-off between task-dependent feature mixing and intrinsic noise. We apply ROME to various reservoir platforms, including spin-wave waveguides and spiking neural networks, demonstrating effective encoder design across physical and neuromorphic reservoirs, even in non-differentiable systems. --- ### HumanOmni-Speaker: Identifying Who said What and When Authors: Detao Bai, Zhiheng Ma, Xihan Wei Published: 2026-03-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.21664v3 While Omni-modal Large Language Models have made strides in joint sensory processing, they fundamentally struggle with a cornerstone of human interaction: deciphering complex, multi-person conversational dynamics to accurately answer ``Who said what and when.'' Current models suffer from an ``illusion of competence'' -- they exploit visual biases in conventional benchmarks to bypass genuine cross-modal alignment, while relying on sparse, low-frame-rate visual sampling that destroys crucial high-frequency dynamics like lip movements. To shatter this illusion, we introduce Visual-Registered Speaker Diarization and Recognition (VR-SDR) and the HumanOmni-Speaker Benchmark. By strictly eliminating visual shortcuts, this rigorous paradigm demands true end-to-end spatio-temporal identity binding using only natural language queries. To overcome the underlying architectural perception gap, we propose HumanOmni-Speaker, powered by a Visual Delta Encoder. By sampling raw video at 25 fps and explicitly compressing inter-frame motion residuals into just 6 tokens per frame, it captures fine-grained visemes and speaker trajectories without triggering a catastrophic token explosion. Ultimately, HumanOmni-Speaker demonstrates strong multimodal synergy, natively enabling end-to-end lip-reading and high-precision spatial localization without intrusive cropping, and achieving superior performance across a wide spectrum of speaker-centric tasks. --- ### TAMTRL: Teacher-Aligned Reward Reshaping for Multi-Turn Reinforcement Learning in Long-Context Compression Authors: Li Wang, Yandong Wang, Xin Yu, Kui Zhang, Tianhao Peng, Wenjun Wu Published: 2026-03-23 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.21663v1 The rapid progress of large language models (LLMs) has led to remarkable performance gains across a wide range of tasks. However, when handling long documents that exceed the model's context window limit, the entire context cannot be processed in a single pass, making chunk-wise processing necessary. This requires multiple turns to read different chunks and update memory. However, supervision is typically provided only by the final outcome, which makes it difficult to evaluate the quality of memory updates at each turn in the multi-turn training setting. This introduces a temporal credit assignment challenge. Existing approaches, such as LLM-as-a-judge or process reward models, incur substantial computational overhead and suffer from estimation noise. To better address the credit assignment problem in multi-turn memory training, we propose Teacher-Aligned Reward Reshaping for Multi-Turn Reinforcement Learning (TAMTRL). TAMTRL leverages relevant documents as teacher signals by aligning them with each turn of model input and assigns rewards through normalized probabilities in a self-supervised manner. This provides fine-grained learning signals for each memory update and improves long-context processing. Experiments with multiple models of varying scales across seven long-context benchmarks show that TAMTRL consistently outperforms strong baselines, demonstrating its effectiveness. Our code is available at https://anonymous.4open.science/r/TAMTRL-F1F8. --- ### Cross-Scenario Deraining Adaptation with Unpaired Data: Superpixel Structural Priors and Multi-Stage Pseudo-Rain Synthesis Authors: Kangbo Zhao, Miaoxin Guan, Xiang Chen, Yukai Shi, Jinshan Pan Published: 2026-03-23 Categories: cs.CV, cs.AI, cs.GR, cs.LG, cs.MM arXiv: https://arxiv.org/abs/2603.21661v1 Image deraining plays a pivotal role in low-level computer vision, serving as a prerequisite for robust outdoor surveillance and autonomous driving systems. While deep learning paradigms have achieved remarkable success in firmly aligned settings, they often suffer from severe performance degradation when generalized to unseen Out-of-Distribution (OOD) scenarios. This failure stems primarily from the significant domain discrepancy between synthetic training datasets and the complex physical dynamics of real-world rain. To address these challenges, this paper proposes a pioneering cross-scenario deraining adaptation framework. Diverging from conventional approaches, our method obviates the requirements for paired rainy observations in the target domain, leveraging exclusively rain-free background images. We design a Superpixel Generation (Sup-Gen) module to extract stable structural priors from the source domain using Simple Linear Iterative Clustering. Subsequently, a Resolution-adaptive Fusion strategy is introduced to align these source structures with target backgrounds through texture similarity, ensuring the synthesis of diverse and realistic pseudo-data. Finally, we implement a pseudo-label re-Synthesize mechanism that employs multi-stage noise generation to simulate realistic rain streaks. This framework functions as a versatile plug-and-play module capable of seamless integration into arbitrary deraining architectures. Extensive experiments on state-of-the-art models demonstrate that our approach yields remarkable PSNR gains of up to 32% to 59% in OOD domains while significantly accelerating training convergence. --- ### OmniFM: Toward Modality-Robust and Task-Agnostic Federated Learning for Heterogeneous Medical Imaging Authors: Meilin Liu, Jiaying Wang, Jing Shan Published: 2026-03-23 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.21660v1 Federated learning (FL) has become a promising paradigm for collaborative medical image analysis, yet existing frameworks remain tightly coupled to task-specific backbones and are fragile under heterogeneous imaging modalities. Such constraints hinder real-world deployment, where institutions vary widely in modality distributions and must support diverse downstream tasks. To address this limitation, we propose OmniFM, a modality- and task-agnostic FL framework that unifies training across classification, segmentation, super-resolution, visual question answering, and multimodal fusion without re-engineering the optimization pipeline. OmniFM builds on a key frequency-domain insight: low-frequency spectral components exhibit strong cross-modality consistency and encode modality-invariant anatomical structures. Accordingly, OmniFM integrates (i) Global Spectral Knowledge Retrieval to inject global frequency priors, (ii) Embedding-wise Cross-Attention Fusion to align representations, and (iii) Prefix-Suffix Spectral Prompting to jointly condition global and personalized cues, together regularized by a Spectral-Proximal Alignment objective that stabilizes aggregation. Experiments on real-world datasets show that OmniFM consistently surpasses state-of-the-art FL baselines across intra- and cross-modality heterogeneity, achieving superior results under both fine-tuning and training-from-scratch setups. --- ### A Comparative Analysis of LLM Memorization at Statistical and Internal Levels: Cross-Model Commonalities and Model-Specific Signatures Authors: Bowen Chen, Namgi Han, Yusuke Miyao Published: 2026-03-23 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2603.21658v1 Memorization is a fundamental component of intelligence for both humans and LLMs. However, while LLM performance scales rapidly, our understanding of memorization lags. Due to limited access to the pre-training data of LLMs, most previous studies focus on a single model series, leading to isolated observations among series, making it unclear which findings are general or specific. In this study, we collect multiple model series (Pythia, OpenLLaMa, StarCoder, OLMo1/2/3) and analyze their shared or unique memorization behavior at both the statistical and internal levels, connecting individual observations while showing new findings. At the statistical level, we reveal that the memorization rate scales log-linearly with model size, and memorized sequences can be further compressed. Further analysis demonstrated a shared frequency and domain distribution pattern for memorized sequences. However, different models also show individual features under the above observations. At the internal level, we find that LLMs can remove certain injected perturbations, while memorized sequences are more sensitive. By decoding middle layers and attention head ablation, we revealed the general decoding process and shared important heads for memorization. However, the distribution of those important heads differs between families, showing a unique family-level feature. Through bridging various experiments and revealing new findings, this study paves the way for a universal and fundamental understanding of memorization in LLM. --- ### TrustFed: Enabling Trustworthy Medical AI under Data Privacy Constraints Authors: Vagish Kumar, Syed Bahauddin Alam, Souvik Chakraborty Published: 2026-03-23 Categories: cs.LG, cs.CY arXiv: https://arxiv.org/abs/2603.21656v1 Protecting patient privacy remains a fundamental barrier to scaling machine learning across healthcare institutions, where centralizing sensitive data is often infeasible due to ethical, legal, and regulatory constraints. Federated learning offers a promising alternative by enabling privacy-preserving, multi-institutional training without sharing raw patient data; however, real-world deployments face severe challenges from data heterogeneity, site-specific biases, and class imbalance, which degrade predictive reliability and render existing uncertainty quantification methods ineffective. Here, we present TrustFed, a federated uncertainty quantification framework that provides distribution-free, finite-sample coverage guarantees under heterogeneous and imbalanced healthcare data, without requiring centralized access. TrustFed introduces a representation-aware client assignment mechanism that leverages internal model representations to enable effective calibration across institutions, along with a soft-nearest threshold aggregation strategy that mitigates assignment uncertainty while producing compact and reliable prediction sets. Using over 430,000 medical images across six clinically distinct imaging modalities, we conduct one of the most comprehensive evaluations of uncertainty-aware federated learning in medical imaging, demonstrating robust coverage guarantees across datasets with diverse class cardinalities and imbalance regimes. By validating TrustFed at this scale and breadth, our study advances uncertainty-aware federated learning from proof-of-concept toward clinically meaningful, modality-agnostic deployment, positioning statistically guaranteed uncertainty as a core requirement for next-generation healthcare AI systems. --- ### Towards Secure Retrieval-Augmented Generation: A Comprehensive Review of Threats, Defenses and Benchmarks Authors: Yanming Mu, Hao Hu, Feiyang Li, Qiao Yuan, Jiang Wu, Zichuan Liu, Pengcheng Liu, Mei Wang, Hongwei Zhou, Yuling Liu Published: 2026-03-23 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2603.21654v1 Retrieval-Augmented Generation (RAG) significantly mitigates the hallucinations and domain knowledge deficiency in large language models by incorporating external knowledge bases. However, the multi-module architecture of RAG introduces complex system-level security vulnerabilities. Guided by the RAG workflow, this paper analyzes the underlying vulnerability mechanisms and systematically categorizes core threat vectors such as data poisoning, adversarial attacks, and membership inference attacks. Based on this threat assessment, we construct a taxonomy of RAG defense technologies from a dual perspective encompassing both input and output stages. The input-side analysis reviews data protection mechanisms including dynamic access control, homomorphic encryption retrieval, and adversarial pre-filtering. The output-side examination summarizes advanced leakage prevention techniques such as federated learning isolation, differential privacy perturbation, and lightweight data sanitization. To establish a unified benchmark for future experimental design, we consolidate authoritative test datasets, security standards, and evaluation frameworks. To the best of our knowledge, this paper presents the first end-to-end survey dedicated to the security of RAG systems. Distinct from existing literature that isolates specific vulnerabilities, we systematically map the entire pipeline-providing a unified analysis of threat models, defense mechanisms, and evaluation benchmarks. By enabling deep insights into potential risks, this work seeks to foster the development of highly robust and trustworthy next-generation RAG systems. --- ### MISApp: Multi-Hop Intent-Aware Session Graph Learning for Next App Prediction Authors: Yunchi Yang, Longlong Li, Jianliang Wu, Cunquan Qu Published: 2026-03-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.21653v2 Predicting the next mobile app a user will launch is essential for proactive mobile services. Yet accurate prediction remains challenging in real-world settings, where user intent can shift rapidly within short sessions and user-specific historical profiles are often sparse or unavailable, especially under cold-start conditions. Existing approaches mainly model app usage as sequential behavior or local session transitions, limiting their ability to capture higher-order structural dependencies and evolving session intent. To address this issue, we propose MISApp, a profile-free framework for next app prediction based on multi-hop session graph learning. MISApp constructs multi-hop session graphs to capture transition dependencies at different structural ranges, learns session representations through lightweight graph propagation, incorporates temporal context and similarity-based spatial categorization to characterize session conditions, and captures intent evolution from recent interactions. Experiments on two real-world app usage datasets show that MISApp consistently outperforms competitive baselines under both standard and cold-start settings, while maintaining a favorable balance between predictive accuracy and practical efficiency. Further analyses show that multi-hop relations capture higher-hop-specific predictive signals beyond direct 1-Hop adjacency, and that the learned hop-level attention weights align well with structural relevance, providing both empirical and interpretable evidence for the effectiveness of the proposed multi-hop modeling strategy. --- ### MISApp: Multi-Hop Intent-Aware Session Graph Learning for Next App Prediction Authors: Yunchi Yang, Longlong Li, Jianliang Wu, Cunquan Qu Published: 2026-03-23 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.21653v1 Predicting the next mobile app a user will launch is essential for proactive mobile services. Yet accurate prediction remains challenging in real-world settings, where user intent can shift rapidly within short sessions and user-specific historical profiles are often sparse or unavailable, especially under cold-start conditions. Existing approaches mainly model app usage as sequential behavior or local session transitions, limiting their ability to capture higher-order structural dependencies and evolving session intent. To address this issue, we propose MISApp, a profile-free framework for next app prediction based on multi-hop session graph learning. MISApp constructs multi-hop session graphs to capture transition dependencies at different structural ranges, learns session representations through lightweight graph propagation, incorporates temporal and spatial context to characterize session conditions, and captures intent evolution from recent interactions. Experiments on two real-world app usage datasets show that MISApp consistently outperforms competitive baselines under both standard and cold-start settings, while maintaining a favorable balance between predictive accuracy and practical efficiency. Further analyses show that the learned hop-level attention weights align well with structural relevance, offering interpretable evidence for the effectiveness of the proposed multi-hop modeling strategy. ---