# Research Papers — 2026-03 · bundle page 171 of 243 > Full-text markdown of papers #8501–8550 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/171 [← Previous bundle](https://aipod.dev/papers/archive/2026-03/page/170.md) [Next bundle →](https://aipod.dev/papers/archive/2026-03/page/172.md) ### An Optimal Control Approach To Transformer Training Authors: Kağan Akman, Naci Saldı, Serdar Yüksel Published: 2026-03-10 Categories: cs.LG, math.OC arXiv: https://arxiv.org/abs/2603.09571v1 In this paper, we develop a rigorous optimal control-theoretic approach to Transformer training that respects key structural constraints such as (i) realized-input-independence during execution, (ii) the ensemble control nature of the problem, and (iii) positional dependence. We model the Transformer architecture as a discrete-time controlled particle system with shared actions, exhibiting noise-free McKean-Vlasov dynamics. While the resulting dynamics is not Markovian, we show that lifting it to probability measures produces a fully-observed Markov decision process (MDP). Positional encodings are incorporated into the state space to preserve the sequence order under lifting. Using the dynamic programming principle, we establish the existence of globally optimal policies under mild assumptions of compactness. We further prove that closed-loop policies in the lifted is equivalent to an initial-distribution dependent open-loop policy, which are realized-input-independent and compatible with standard Transformer training. To train a Transformer, we propose a triply quantized training procedure for the lifted MDP by quantizing the state space, the space of probability measures, and the action space, and show that any optimal policy for the triply quantized model is near-optimal for the original training problem. Finally, we establish stability and empirical consistency properties of the lifted model by showing that the value function is continuous with respect to the perturbations of the initial empirical measures and convergence of policies as the data size increases. This approach provides a globally optimal and robust alternative to gradient-based training without requiring smoothness or convexity. --- ### GeoAlignCLIP: Enhancing Fine-Grained Vision-Language Alignment in Remote Sensing via Multi-Granular Consistency Learning Authors: Xiao Yang, Ronghao Fu, Zhuoran Duan, Zhiwen Lin, Xueyan Liu, Bo Yang Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09566v1 Vision-language pretraining models have made significant progress in bridging remote sensing imagery with natural language. However, existing approaches often fail to effectively integrate multi-granular visual and textual information, relying primarily on global image-text alignment. This limitation hinders the model's ability to accurately capture fine-grained details in images, thus restricting its performance in complex, fine-grained tasks. To address this, we propose GeoAlignCLIP, a unified framework that achieves fine-grained alignment in remote sensing tasks by learning multi-granular semantic alignments and incorporating intra-modal consistency, enabling more precise visual-semantic alignment between image regions and text concepts. Additionally, we construct RSFG-100k, a fine-granular remote sensing dataset containing scene descriptions, region-level annotations, and challenging hard-negative samples, providing hierarchical supervision for model training. Extensive experiments conducted on multiple public remote-sensing benchmarks demonstrate that GeoAlignCLIP consistently outperforms existing RS-specific methods across diverse tasks, exhibiting more robust and accurate fine-grained vision-language alignment. --- ### a-TMFG: Scalable Triangulated Maximally Filtered Graphs via Approximate Nearest Neighbors Authors: Lionel Yelibi Published: 2026-03-10 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2603.09564v1 The traditional Triangular Maximally Filtered Graph (TMFG) construction requires pre-computation and storage of a dense correlation matrix; this limits its applicability to small and medium-sized datasets. Here we identify key memory and runtime complexity challenges when using TMFG at scale. We then present the Approximate Triangular Maximally Filtered Graph (a-TMFG) algorithm. This is a novel approach to scaling the construction of artificial graphs from data inspired by TMFG. The method employs k-Nearest Neighbors Graphs (kNNG) for initial construction, and implements a memory management strategy to search and estimate missing correlations on-the-fly. This provides representations to control combinatorial explosion. The algorithm is tested for robustness to the parameters and noise, and is evaluated on datasets with millions of observations. This new method provides a parsimonious way to construct graphs for use-cases where graphs are used as input to supervised and unsupervised learning but where no natural graph exists. --- ### Learning Bayesian and Markov Networks with an Unreliable Oracle Authors: Juha Harviainen, Pekka Parviainen, Vidya Sagar Sharma Published: 2026-03-10 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.09563v1 We study constraint-based structure learning of Markov networks and Bayesian networks in the presence of an unreliable conditional independence oracle that makes at most a bounded number of errors. For Markov networks, we observe that a low maximum number of vertex-wise disjoint paths implies that the structure is uniquely identifiable even if the number of errors is (moderately) exponential in the number of vertices. For Bayesian networks, however, we prove that one cannot tolerate any errors to always identify the structure even when many commonly used graph parameters like treewidth are bounded. Finally, we give algorithms for structure learning when the structure is uniquely identifiable. --- ### ALARM: Audio-Language Alignment for Reasoning Models Authors: Petr Grinberg, Hassan Shahmohammadi Published: 2026-03-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.09556v1 Large audio language models (ALMs) extend LLMs with auditory understanding. A common approach freezes the LLM and trains only an adapter on self-generated targets. However, this fails for reasoning LLMs (RLMs) whose built-in chain-of-thought traces expose the textual surrogate input, yielding unnatural responses. We propose self-rephrasing, converting self-generated responses into audio-understanding variants compatible with RLMs while preserving distributional alignment. We further fuse and compress multiple audio encoders for stronger representations. For training, we construct a 6M-instance multi-task corpus (2.5M unique prompts) spanning 19K hours of speech, music, and sound. Our 4B-parameter ALM outperforms similarly sized models and surpasses most larger ALMs on related audio-reasoning benchmarks, while preserving textual capabilities with a low training cost. Notably, we achieve the best open-source result on the MMAU-speech and MMSU benchmarks and rank third among all the models. --- ### Compiler-First State Space Duality and Portable $O(1)$ Autoregressive Caching for Inference Authors: Cosmo Santoni, Anmol Thapar Published: 2026-03-10 Categories: cs.LG, cs.AI, cs.DC, cs.PF arXiv: https://arxiv.org/abs/2603.09555v2 High-throughput Mamba-2 inference is usually tied to fused CUDA and Triton kernels, limiting portability across accelerator backends. We show that the state space duality (SSD) recurrence has a compiler-friendly structure: diagonal per-head dynamics, fixed-size chunking, einsum-dominated compute, and static control flow. Expressing this structure in standard JAX primitives gives a single-source inference path with no custom kernels, a registered JAX PyTree cache, and a compiled on-device autoregressive loop. On a single Google Cloud TPU v6e, batch-1 prefill reaches approximately 140 TFLOPS, or 15% model FLOP utilisation (MFU), the roofline ceiling for this regime, and cached decode reaches up to 64% hardware bandwidth utilisation (HBU). At a 4096-token context, cached decode is 27x--36x faster than full-prefix recomputation across five Mamba-2 checkpoints from 130M to 2.7B parameters. The same source runs unmodified on NVIDIA L40S, where cached decode remains sequence-length independent across all model scales. WikiText-103 validation perplexity matches the Triton reference mamba_ssm v2.2.2 within +/-0.0005 points, and hidden states agree to float32 rounding tolerance. Code is available at https://github.com/CosmoNaught/mamba2-jax. --- ### GeoSolver: Scaling Test-Time Reasoning in Remote Sensing with Fine-Grained Process Supervision Authors: Lang Sun, Ronghao Fu, Zhuoran Duan, Haoran Liu, Xueyan Liu, Bo Yang Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09551v2 While Vision-Language Models (VLMs) have significantly advanced remote sensing interpretation, enabling them to perform complex, step-by-step reasoning remains highly challenging. Recent efforts to introduce Chain-of-Thought (CoT) reasoning to this domain have shown promise, yet ensuring the visual faithfulness of these intermediate steps remains a critical bottleneck. To address this, we introduce GeoSolver, a novel framework that transitions remote sensing reasoning toward verifiable, process-supervised reinforcement learning. We first construct Geo-PRM-2M, a large-scale, token-level process supervision dataset synthesized via entropy-guided Monte Carlo Tree Search (MCTS) and targeted visual hallucination injection. Building upon this dataset, we train GeoPRM, a token-level process reward model (PRM) that provides granular faithfulness feedback. To effectively leverage these verification signals, we propose Process-Aware Tree-GRPO, a reinforcement learning algorithm that integrates tree-structured exploration with a faithfulness-weighted reward mechanism to precisely assign credit to intermediate steps. Extensive experiments demonstrate that our resulting model, GeoSolver-9B, achieves state-of-the-art performance across diverse remote sensing benchmarks. Crucially, GeoPRM unlocks robust Test-Time Scaling (TTS). Serving as a universal geospatial verifier, it seamlessly scales the performance of GeoSolver-9B and directly enhances general-purpose VLMs, highlighting its remarkable cross-model generalization. --- ### A comprehensive study of time-of-flight non-line-of-sight imaging Authors: Julio Marco, Adrian Jarabo, Ji Hyun Nam, Alberto Tosi, Diego Gutierrez, Andreas Velten Published: 2026-03-10 Categories: cs.CV, cs.GR arXiv: https://arxiv.org/abs/2603.09548v1 Time-of-Flight non-line-of-sight (ToF NLOS) imaging techniques provide state-of-the-art reconstructions of scenes hidden around corners by inverting the optical path of indirect photons scattered by visible surfaces and measured by picosecond resolution sensors. The emergence of a wide range of ToF NLOS imaging methods with heterogeneous formulae and hardware implementations obscures the assessment of both their theoretical and experimental aspects. We present a comprehensive study of a representative set of ToF NLOS imaging methods by discussing their similarities and differences under common formulation and hardware. We first outline the problem statement under a common general forward model for ToF NLOS measurements, and the typical assumptions that yield tractable inverse models. We discuss the relationship of the resulting simplified forward and inverse models to a family of Radon transforms, and how migrating these to the frequency domain relates to recent phasor-based virtual line-of-sight imaging models for NLOS imaging that obey the constraints of conventional lens-based imaging systems. We then evaluate performance of the selected methods on hidden scenes captured under the same hardware setup and similar photon counts. Our experiments show that existing methods share similar limitations on spatial resolution, visibility, and sensitivity to noise when operating under equal hardware constraints, with particular differences that stem from method-specific parameters. We expect our methodology to become a reference in future research on ToF NLOS imaging to obtain objective comparisons of existing and new methods. --- ### Memory-Guided View Refinement for Dynamic Human-in-the-loop EQA Authors: Xin Lu, Rui Li, Xun Huang, Weixin Li, Chuanqing Zhuang, Jiayuan Li, Zhengda Lu, Jun Xiao, Yunhong Wang Published: 2026-03-10 Categories: cs.CV, cs.MM arXiv: https://arxiv.org/abs/2603.09541v1 Embodied Question Answering (EQA) has traditionally been evaluated in temporally stable environments where visual evidence can be accumulated reliably. However, in dynamic, human-populated scenes, human activities and occlusions introduce significant perceptual non-stationarity: task-relevant cues are transient and view-dependent, while a store-then-retrieve strategy over-accumulates redundant evidence and increases inference cost. This setting exposes two practical challenges for EQA agents: resolving ambiguity caused by viewpoint-dependent occlusions, and maintaining compact yet up-to-date evidence for efficient inference. To enable systematic study of this setting, we introduce DynHiL-EQA, a human-in-the-loop EQA dataset with two subsets: a Dynamic subset featuring human activities and temporal changes, and a Static subset with temporally stable observations. To address the above challenges, we present DIVRR (Dynamic-Informed View Refinement and Relevance-guided Adaptive Memory Selection), a training-free framework that couples relevance-guided view refinement with selective memory admission. By verifying ambiguous observations before committing them and retaining only informative evidence, DIVRR improves robustness under occlusions while preserving fast inference with compact memory. Extensive experiments on DynHiL-EQA and the established HM-EQA dataset demonstrate that DIVRR consistently improves over existing baselines in both dynamic and static settings while maintaining high inference efficiency. --- ### Towards Unified Multimodal Interleaved Generation via Group Relative Policy Optimization Authors: Ming Nie, Chunwei Wang, Jianhua Han, Hang Xu, Li Zhang Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09538v1 Unified vision-language models have made significant progress in multimodal understanding and generation, yet they largely fall short in producing multimodal interleaved outputs, which is a crucial capability for tasks like visual storytelling and step-by-step visual reasoning. In this work, we propose a reinforcement learning-based post-training strategy to unlock this capability in existing unified models, without relying on large-scale multimodal interleaved datasets. We begin with a warm-up stage using a hybrid dataset comprising curated interleaved sequences and limited data for multimodal understanding and text-to-image generation, which exposes the model to interleaved generation patterns while preserving its pretrained capabilities. To further refine interleaved generation, we propose a unified policy optimization framework that extends Group Relative Policy Optimization (GRPO) to the multimodal setting. Our approach jointly models text and image generation within a single decoding trajectory and optimizes it with our novel hybrid rewards covering textual relevance, visual-text alignment, and structural fidelity. Additionally, we incorporate process-level rewards to provide step-wise guidance, enhancing training efficiency in complex multimodal tasks. Experiments on MMIE and InterleavedBench demonstrate that our approach significantly enhances the quality and coherence of multimodal interleaved generation. --- ### Enhancing Debunking Effectiveness through LLM-based Personality Adaptation Authors: Pietro Dell'Oglio, Alessandro Bondielli, Francesco Marcelloni, Lucia C. Passaro Published: 2026-03-10 Categories: cs.AI, cs.CL arXiv: https://arxiv.org/abs/2603.09533v1 This study proposes a novel methodology for generating personalized fake news debunking messages by prompting Large Language Models (LLMs) with persona-based inputs aligned to the Big Five personality traits: Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness. Our approach guides LLMs to transform generic debunking content into personalized versions tailored to specific personality profiles. To assess the effectiveness of these transformations, we employ a separate LLM as an automated evaluator simulating corresponding personality traits, thereby eliminating the need for costly human evaluation panels. Our results show that personalized messages are generally seen as more persuasive than generic ones. We also find that traits like Openness tend to increase persuadability, while Neuroticism can lower it. Differences between LLM evaluators suggest that using multiple models provides a clearer picture. Overall, this work demonstrates a practical way to create more targeted debunking messages exploiting LLMs, while also raising important ethical questions about how such technology might be used. --- ### What Do We Care About in Bandits with Noncompliance? BRACE: Bandits with Recommendations, Abstention, and Certified Effects Authors: Nicolás Della Penna Published: 2026-03-10 Categories: stat.ML, cs.LG arXiv: https://arxiv.org/abs/2603.09532v1 Bandits with noncompliance separate the learner's recommendation from the treatment actually delivered, so the learning target itself must be chosen. A platform may care about recommendation welfare in the current mediated workflow, treatment learning for a future direct-control regime, or anytime-valid uncertainty for one of those targets. These objectives need not agree. We formalize this objective-choice problem, identify the direct-control regime in which recommendation and treatment objectives collapse, and show by example that recommendation welfare can strictly exceed every learner-measurable treatment policy when downstream actors use private information. For finite-context square-IV problems we propose BRACE, a parameter-free phase-doubling algorithm that performs IV inversion only after matrix certification and otherwise returns full-range but honest structural intervals. BRACE delivers simultaneous policy-value validity, fixed-gap identification of the operationally optimal recommendation policy, and fixed-gap identification of the structurally optimal treatment policy under contextual homogeneity and invertibility. We complement the theory with a finite-context empirical benchmark spanning direct control, mediated present-versus-future tradeoffs, weak identification, homogeneity failure, and rectangular overidentification. The experiments show that safety appears as regret on easy problems, as abstention and wide valid intervals under weak identification, as a reason to prefer recommendation welfare under homogeneity failure, and as tighter structural uncertainty when extra instruments are available. For rich contexts, we also derive an orthogonal score whose conditional bias factorizes into compliance-model and outcome-model errors, clarifying what must be stabilized for anytime-valid semiparametric IV inference. --- ### Association of Progressive PPFE and Mortality in Lung Cancer Screening Cohorts Authors: Shahab Aslani, Mehran Azimbagirad, Daryl Cheng, Daisuke Yamada, Ryoko Egashira, Adam Szmul, Justine Chan-Fook, Robert Chapman, Alfred Chung Pui So, Shanshan Wang, John McCabe, Tianqi Yang et al. Published: 2026-03-10 Categories: q-bio.QM, cs.CV, eess.IV, stat.AP arXiv: https://arxiv.org/abs/2603.09531v2 Background: Pleuroparenchymal fibroelastosis (PPFE) is an upper lobe predominant fibrotic lung abnormality associated with increased mortality in established interstitial lung disease. However, the clinical significance of radiologic PPFE progression in lung cancer screening (LCS) populations remains unclear. Methods: We analysed longitudinal low-dose CT scans and clinical data from two LCS studies: National Lung Screening Trial (NLST; n=7,980); SUMMIT study (n=8,561). An automated algorithm quantified PPFE volume on baseline and follow-up scans. Annualised change in PPFE was derived and dichotomised using a distribution-based threshold to define progressive PPFE. Associations between progressive PPFE and mortality were evaluated using Cox proportional hazards models adjusted for demographic and clinical variables. In SUMMIT cohort, associations between progressive PPFE and clinical outcomes were assessed using incidence rate ratios (IRR) and odds ratios (OR). Findings: Progressive PPFE independently associated with mortality in both LCS cohorts (NLST: Hazard Ratio (HR)=1.25, 95% Confidence Interval (CI): 1.01--1.56, p=0.042; SUMMIT: HR=3.14, 95% CI: 1.66--5.97, p<0.001). Within SUMMIT, progressive PPFE was strongly associated with higher respiratory admissions (IRR=2.79, p<0.001), increased antibiotic and steroid use (IRR=1.55, p=0.010), and showed a trend towards higher modified medical research council scores (OR=1.40, p=0.055). Interpretation: Radiologic PPFE progression independently associates with mortality across two large LCS cohorts, and associates with adverse clinical outcomes. Quantitative assessment of PPFE progression may provide a clinically relevant imaging biomarker to identify individuals at increased risk of respiratory morbidity within LCS programmes. --- ### DCAU-Net: Differential Cross Attention and Channel-Spatial Feature Fusion for Medical Image Segmentation Authors: Yanxin Li, Hui Wan, Libin Lan Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09530v1 Accurate medical image segmentation requires effective modeling of both long-range dependencies and fine-grained boundary details. While transformers mitigate the issue of insufficient semantic information arising from the limited receptive field inherent in convolutional neural networks, they introduce new challenges: standard self-attention incurs quadratic computational complexity and often assigns non-negligible attention weights to irrelevant regions, diluting focus on discriminative structures and ultimately compromising segmentation accuracy. Existing attention variants, although effective in reducing computational complexity, fail to suppress redundant computation and inadvertently impair global context modeling. Furthermore, conventional fusion strategies in encoder-decoder architectures, typically based on simple concatenation or summation, can not adaptively integrate high-level semantic information with low-level spatial details. To address these limitations, we propose DCAU-Net, a novel yet efficient segmentation framework with two key ideas. First, a new Differential Cross Attention (DCA) is designed to compute the difference between two independent softmax attention maps to adaptively highlight discriminative structures. By replacing pixel-wise key and value tokens with window-level summary tokens, DCA dramatically reduces computational complexity without sacrificing precision. Second, a Channel-Spatial Feature Fusion (CSFF) strategy is introduced to adaptively recalibrate features from skip connections and up-sampling paths through using sequential channel and spatial attention, effectively suppressing redundant information and amplifying salient cues. Experiments on two public benchmarks demonstrate that DCAU-Net achieves competitive performance with enhanced segmentation accuracy and robustness. --- ### RESBev: Making BEV Perception More Robust Authors: Lifeng Zhuo, Kefan Jin, Zhe Liu, Hesheng Wang Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09529v2 Bird's-eye-view (BEV) perception has emerged as a cornerstone of autonomous driving systems, providing a structured, ego-centric representation critical for downstream planning and control. However, real-world deployment faces challenges from sensor degradation and adversarial attacks, which can cause severe perceptual anomalies and ultimately compromise the safety of autonomous driving systems. To address this, we propose a resilient and plug-and-play BEV perception method, RESBev, which can be easily applied to existing BEV perception methods to enhance their robustness to diverse disturbances. Specifically, we reframe perception robustness as a latent semantic prediction problem. A latent world model is constructed to extract spatiotemporal correlations across sequential BEV observations, thereby learning the underlying BEV state transitions to predict clean BEV features for reconstructing corrupted observations. The proposed framework operates at the semantic feature level of the Lift-Splat-Shoot pipeline, enabling recovery that generalizes across both natural disturbances and adversarial attacks without modifying the underlying backbone. Extensive experiments on the nuScenes dataset demonstrate that, with few-shot fine-tuning, RESBev significantly improves the robustness of existing BEV perception models against various external disturbances and adversarial attacks. --- ### Efficiently Aligning Draft Models via Parameter- and Data-Efficient Adaptation Authors: Luxi Lin, Zhihang Lin, Zhanpeng Zeng, Yuhao Chen, Qingyu Zhang, Jixiang Luo, Xuelong Li, Rongrong Ji Published: 2026-03-10 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2603.09527v1 Speculative decoding accelerates LLM inference but suffers from performance degradation when target models are fine-tuned for specific domains. A naive solution is to retrain draft models for every target model, which is costly and inefficient. To address this, we introduce a parameter- and data-efficient framework named Efficient Draft Adaptation, abbreviated as EDA, for efficiently adapting draft models. EDA introduces three innovations: (1) a decoupled architecture that utilizes shared and private components to model the shared and target-specific output distributions separately, enabling parameter-efficient adaptation by updating only the lightweight private component;(2) a data regeneration strategy that utilizes the fine-tuned target model to regenerate training data, thereby improving the alignment between training and speculative decoding, leading to higher average acceptance length;(3) a sample selection mechanism that prioritizes high-value data for efficient adaptation. Our experiments show that EDA effectively restores speculative performance on fine-tuned models, achieving superior average acceptance lengths with significantly reduced training costs compared to full retraining. Code is available at https://github.com/Lyn-Lucy/Efficient-Draft-Adaptation. --- ### You Didn't Have to Say It like That: Subliminal Learning from Faithful Paraphrases Authors: Isaia Gisler, Zhonghao He, Tianyi Qiu Published: 2026-03-10 Categories: cs.CL, cs.LG arXiv: https://arxiv.org/abs/2603.09517v1 When language models are trained on synthetic data, they (student model) can covertly acquire behavioral traits from the data-generating model (teacher model). Subliminal learning refers to the transmission of traits from a teacher to a student model via training on data unrelated to those traits. Prior work demonstrated this in the training domains of number sequences, code, and math Chain-of-Thought traces including transmission of misaligned behaviors. We investigate whether transmission occurs through natural language paraphrases with fixed semantic content, and whether content explicitly contradicting the teacher's preference can block it. We find that training on paraphrases from a teacher system-prompted to love a particular animal increases a student's preference for that animal by up to 19 percentage points. This occurs when paraphrased content is semantically unrelated to the animal, or even when it explicitly expresses dislike. The transmission succeeds despite aggressive filtering to ensure paraphrase fidelity. This raises concerns for pipelines where models generate their own training data: content-based inspection cannot detect such transmission, and even preference-contradicting content fails to prevent it. --- ### Probing the Reliability of Driving VLMs: From Inconsistent Responses to Grounded Temporal Reasoning Authors: Chun-Peng Chang, Chen-Yu Wang, Holger Caesar, Alain Pagani Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09512v1 A reliable driving assistant should provide consistent responses based on temporally grounded reasoning derived from observed information. In this work, we investigate whether Vision-Language Models (VLMs), when applied as driving assistants, can response consistantly and understand how present observations shape future outcomes, or whether their outputs merely reflect patterns memorized during training without temporally grounded reasoning. While recent efforts have integrated VLMs into autonomous driving, prior studies typically emphasize scene understanding and instruction generation, implicitly assuming that strong visual interpretation naturally enables consistant future reasoning and thus ensures reliable decision-making, a claim we critically examine. We focus on two major challenges limiting VLM reliability in this setting: response inconsistency, where minor input perturbations yield different answers or, in some cases, responses degenerate toward near-random guessing, and limited temporal reasoning, in which models fail to reason and align sequential events from current observations, often resulting in incorrect or even contradictory responses. Moreover, we find that models with strong visual understanding do not necessarily perform best on tasks requiring temporal reasoning, indicating a tendency to over-rely on pretrained patterns rather than modeling temporal dynamics. To address these issues, we adopt existing evaluation methods and introduce FutureVQA, a human-annotated benchmark dataset specifically designed to assess future scene reasoning. In addition, we propose a simple yet effective self-supervised tuning approach with chain-of-thought reasoning that improves both consistency and temporal reasoning without requiring temporal labels. --- ### TrainDeeploy: Hardware-Accelerated Parameter-Efficient Fine-Tuning of Small Transformer Models at the Extreme Edge Authors: Run Wang, Victor J. B. Jung, Philip Wiese, Francesco Conti, Alessio Burrello, Luca Benini Published: 2026-03-10 Categories: cs.AR, cs.LG arXiv: https://arxiv.org/abs/2603.09511v1 On-device tuning of deep neural networks enables long-term adaptation at the edge while preserving data privacy. However, the high computational and memory demands of backpropagation pose significant challenges for ultra-low-power, memory-constrained extreme-edge devices. These challenges are further amplified for attention-based models due to their architectural complexity and computational scale. We present TrainDeeploy, a framework that unifies efficient inference and on-device training on heterogeneous ultra-low-power System-on-Chips (SoCs). TrainDeeploy provides the first complete on-device training pipeline for extreme-edge SoCs supporting both Convolutional Neural Networks (CNNs) and Transformer models, together with multiple training strategies such as selective layer-wise fine-tuning and Low-Rank Adaptation (LoRA). On a RISC-V-based heterogeneous SoC, we demonstrate the first end-to-end on-device fine-tuning of a Compact Convolutional Transformer (CCT), achieving up to 11 trained images per second. We show that LoRA reduces dynamic memory usage by 23%, decreases the number of trainable parameters and gradients by 15x, and reduces memory transfer volume by 1.6x compared to full backpropagation. TrainDeeploy achieves up to 4.6 FLOP/cycle on CCT (0.28M parameters, 71-126M FLOPs) and up to 13.4 FLOP/cycle on Deep-AE (0.27M parameters, 0.8M FLOPs), while expanding the scope of prior frameworks to support both CNN and Transformer models with parameter-efficient tuning on extreme-edge platforms. --- ### Context-Nav: Context-Driven Exploration and Viewpoint-Aware 3D Spatial Reasoning for Instance Navigation Authors: Won Shik Jang, Ue-Hwan Kim Published: 2026-03-10 Categories: cs.CV, cs.RO arXiv: https://arxiv.org/abs/2603.09506v3 Text-goal instance navigation (TGIN) asks an agent to resolve a single, free-form description into actions that reach the correct object instance among same-category distractors. We present \textit{Context-Nav}, which elevates long, contextual captions from a local matching cue to a global exploration prior and verifies candidates through 3D spatial reasoning. First, we compute dense text-image alignments for a value map that ranks frontiers -- guiding exploration toward regions consistent with the entire description rather than early detections. Second, upon observing a candidate, we perform a viewpoint-aware relation check: the agent samples plausible observer poses, aligns local frames, and accepts a target only if the spatial relations can be satisfied from at least one viewpoint. The pipeline requires no task-specific training or fine-tuning; we attain state-of-the-art performance on InstanceNav and CoIN-Bench. Ablations show that (i) encoding full captions into the value map avoids wasted motion and (ii) explicit, viewpoint-aware 3D verification prevents semantically plausible but incorrect stops. This suggests that geometry-grounded spatial reasoning is a scalable alternative to heavy policy training or human-in-the-loop interaction for fine-grained instance disambiguation in cluttered 3D scenes. --- ### Modelling the Diachronic Emergence of Phoneme Frequency Distributions Authors: Fermín Moscoso del Prado Martín, Suchir Salhan Published: 2026-03-10 Categories: cs.CL arXiv: https://arxiv.org/abs/2603.09503v1 Phoneme frequency distributions exhibit robust statistical regularities across languages, including exponential-tailed rank-frequency patterns and a negative relationship between phonemic inventory size and the relative entropy of the distribution. The origin of these patterns remains largely unexplained. In this paper, we investigate whether they can arise as consequences of the historical processes that shape phonological systems. We introduce a stochastic model of phonological change and simulate the diachronic evolution of phoneme inventories. A naïve version of the model reproduces the general shape of phoneme rank-frequency distributions but fails to capture other empirical properties. Extending the model with two additional assumptions -- an effect related to functional load and a stabilising tendency toward a preferred inventory size -- yields simulations that match both the observed distributions and the negative relationship between inventory size and relative entropy. These results suggest that some statistical regularities of phonological systems may arise as natural consequences of diachronic sound change rather than from explicit optimisation or compensatory mechanisms. --- ### SurgFed: Language-guided Multi-Task Federated Learning for Surgical Video Understanding Authors: Zheng Fang, Ziwei Niu, Ziyue Wang, Zhu Zhuo, Haofeng Liu, Shuyang Qian, Jun Xia, Yueming Jin Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09496v1 Surgical scene Multi-Task Federated Learning (MTFL) is essential for robot-assisted minimally invasive surgery (RAS) but remains underexplored in surgical video understanding due to two key challenges: (1) Tissue Diversity: Local models struggle to adapt to site-specific tissue features, limiting their effectiveness in heterogeneous clinical environments and leading to poor local predictions. (2) Task Diversity: Server-side aggregation, relying solely on gradient-based clustering, often produces suboptimal or incorrect parameter updates due to inter-site task heterogeneity, resulting in inaccurate localization. In light of these two issues, we propose SurgFed, a multi-task federated learning framework, enabling federated learning for surgical scene segmentation and depth estimation across diverse surgical types. SurgFed is powered by two appealing designs, i.e., Language-guided Channel Selection (LCS) and Language-guided Hyper Aggregation (LHA), to address the challenge of fully exploration on corss-site and cross-task. Technically, the LCS is first designed a lightweight personalized channel selection network that enhances site-specific adaptation using pre-defined text inputs, which optimally the local model learn the specific embeddings. We further introduce the LHA that employs a layer-wise cross-attention mechanism with pre-defined text inputs to model task interactions across sites and guide a hypernetwork for personalized parameter updates. Extensive empirical evidence shows that SurgFed yields improvements over the state-of-the-art methods in five public datasets across four surgical types. The code is available at https://anonymous.4open.science/r/SurgFed-070E/. --- ### Guided Prompt Evolution for Vision-Language Models Adaptation Authors: Enming Zhang, Jiayang Li, Yanlong Wang, Yanru Wu, Zhenyu Liu, Yang Li Published: 2026-03-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2603.09493v3 The adaptation of large-scale vision-language models (VLMs) to downstream tasks with limited labeled data remains a significant challenge. While parameter-efficient prompt learning methods offer a promising path, they often suffer from catastrophic forgetting of pre-trained knowledge. Toward addressing this limitation, our work is grounded in the insight that governing the evolutionary path of prompts is essential for forgetting-free adaptation. To this end, we propose EvoPrompt, a novel framework designed to explicitly steer the prompt trajectory for knowledge-preserving fine-tuning. Specifically, our approach employs a Modality-Shared Prompt Projector (MPP) to generate hierarchical prompts from a unified embedding space. Critically, an evolutionary training strategy decouples low-rank updates into directional and magnitude components, preserving early-learned semantic directions while only adapting their magnitude, thus enabling prompts to evolve without discarding foundational knowledge. Extensive experiments demonstrate that EvoPrompt achieves state-of-the-art performance in few-shot learning while robustly preserving the original zero-shot capabilities of pre-trained VLMs. --- ### EvoPrompt: Guided Prompt Evolution for Vision-Language Models Adaptation Authors: Enming Zhang, Jiayang Li, Yanlong Wang, Yanru Wu, Zhenyu Liu, Yang Li Published: 2026-03-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2603.09493v2 The adaptation of large-scale vision-language models (VLMs) to downstream tasks with limited labeled data remains a significant challenge. While parameter-efficient prompt learning methods offer a promising path, they often suffer from catastrophic forgetting of pre-trained knowledge. Toward addressing this limitation, our work is grounded in the insight that governing the evolutionary path of prompts is essential for forgetting-free adaptation. To this end, we propose EvoPrompt, a novel framework designed to explicitly steer the prompt trajectory for knowledge-preserving fine-tuning. Specifically, our approach employs a Modality-Shared Prompt Projector (MPP) to generate hierarchical prompts from a unified embedding space. Critically, an evolutionary training strategy decouples low-rank updates into directional and magnitude components, preserving early-learned semantic directions while only adapting their magnitude, thus enabling prompts to evolve without discarding foundational knowledge. This process is further stabilized by Feature Geometric Regularization (FGR), which enforces feature decorrelation to prevent representation collapse. Extensive experiments demonstrate that EvoPrompt achieves state-of-the-art performance in few-shot learning while robustly preserving the original zero-shot capabilities of pre-trained VLMs. --- ### Temporal-Conditioned Normalizing Flows for Multivariate Time Series Anomaly Detection Authors: David Baumgartner, Helge Langseth, Kenth Engø-Monsen, Heri Ramampiaro Published: 2026-03-10 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2603.09490v1 This paper introduces temporal-conditioned normalizing flows (tcNF), a novel framework that addresses anomaly detection in time series data with accurate modeling of temporal dependencies and uncertainty. By conditioning normalizing flows on previous observations, tcNF effectively captures complex temporal dynamics and generates accurate probability distributions of expected behavior. This autoregressive approach enables robust anomaly detection by identifying low-probability events within the learned distribution. We evaluate tcNF on diverse datasets, demonstrating good accuracy and robustness compared to existing methods. A comprehensive analysis of strengths and limitations and open-source code is provided to facilitate reproducibility and future research. --- ### Streaming Autoregressive Video Generation via Diagonal Distillation Authors: Jinxiu Liu, Xuanming Liu, Kangfu Mei, Yandong Wen, Ming-Hsuan Yang, Weiyang Liu Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09488v2 Large pretrained diffusion models have significantly enhanced the quality of generated videos, and yet their use in real-time streaming remains limited. Autoregressive models offer a natural framework for sequential frame synthesis but require heavy computation to achieve high fidelity. Diffusion distillation can compress these models into efficient few-step variants, but existing video distillation approaches largely adapt image-specific methods that neglect temporal dependencies. These techniques often excel in image generation but underperform in video synthesis, exhibiting reduced motion coherence, error accumulation over long sequences, and a latency-quality trade-off. We identify two factors that result in these limitations: insufficient utilization of temporal context during step reduction and implicit prediction of subsequent noise levels in next-chunk prediction (i.e., exposure bias). To address these issues, we propose Diagonal Distillation, which operates orthogonally to existing approaches and better exploits temporal information across both video chunks and denoising steps. Central to our approach is an asymmetric generation strategy: more steps early, fewer steps later. This design allows later chunks to inherit rich appearance information from thoroughly processed early chunks, while using partially denoised chunks as conditional inputs for subsequent synthesis. By aligning the implicit prediction of subsequent noise levels during chunk generation with the actual inference conditions, our approach mitigates error propagation and reduces oversaturation in long-range sequences. We further incorporate implicit optical flow modeling to preserve motion quality under strict step constraints. Our method generates a 5-second video in 2.61 seconds (up to 31 FPS), achieving a 277.3x speedup over the undistilled model. --- ### Vibe-Creation: The Epistemology of Human-AI Emergent Cognition Authors: Ilya Levin Published: 2026-03-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.09486v1 The encounter between human reasoning and generative artificial intelligence (GenAI) cannot be adequately described by inherited metaphors of tool use, augmentation, or collaborative partnership. This article argues that such interactions produce a qualitatively distinct cognitive-epistemic formation, designated here as the Third Entity: an emergent, transient structure that arises from the transductive coupling of two ontologically incommensurable modes of cognition. Drawing on Peirce semiotics, Polanyi theory of tacit knowledge, Simondon philosophy of individuation, Ihde postphenomenology, and Morin complexity theory, we develop a multi-layered theoretical account of this formation. We introduce the concept of vibe-creation to designate the pre-reflective cognitive mode through which the Third Entity navigates high-dimensional semantic space and argue that this mode constitutes the automation of tacit knowledge - a development with far-reaching consequences for epistemology, the philosophy of mind, and educational theory. We further propose the notion of asymmetric emergence to characterize the agency of the Third Entity: genuinely novel and irreducible, yet anchored in human intentional responsibility. The article concludes by examining the implications of this theoretical framework for the transformation of educational institutions and the redefinition of intellectual competence in the age of GenAI. --- ### Component-Aware Sketch-to-Image Generation Using Self-Attention Encoding and Coordinate-Preserving Fusion Authors: Ali Zia, Muhammad Umer Ramzan, Usman Ali, Muhammad Faheem, Abdelwahed Khamis, Shahnawaz Qureshi Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09484v1 Translating freehand sketches into photorealistic images remains a fundamental challenge in image synthesis, particularly due to the abstract, sparse, and stylistically diverse nature of sketches. Existing approaches, including GAN-based and diffusion-based models, often struggle to reconstruct fine-grained details, maintain spatial alignment, or adapt across different sketch domains. In this paper, we propose a component-aware, self-refining framework for sketch-to-image generation that addresses these challenges through a novel two-stage architecture. A Self-Attention-based Autoencoder Network (SA2N) first captures localised semantic and structural features from component-wise sketch regions, while a Coordinate-Preserving Gated Fusion (CGF) module integrates these into a coherent spatial layout. Finally, a Spatially Adaptive Refinement Revisor (SARR), built on a modified StyleGAN2 backbone, enhances realism and consistency through iterative refinement guided by spatial context. Extensive experiments across both facial (CelebAMask-HQ, CUFSF) and non-facial (Sketchy, ChairsV2, ShoesV2) datasets demonstrate the robustness and generalizability of our method. The proposed framework consistently outperforms state-of-the-art GAN and diffusion models, achieving significant gains in image fidelity, semantic accuracy, and perceptual quality. On CelebAMask-HQ, our model improves over prior methods by 21% (FID), 58% (IS), 41% (KID), and 20% (SSIM). These results, along with higher efficiency and visual coherence across diverse domains, position our approach as a strong candidate for applications in forensics, digital art restoration, and general sketch-based image synthesis. --- ### GenePlan: Evolving Better Generalized PDDL Plans using Large Language Models Authors: Andrew Murray, Danial Dervovic, Alberto Pozanco, Michael Cashmore Published: 2026-03-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.09481v1 We present GenePlan (GENeralized Evolutionary Planner), a novel framework that leverages large language model (LLM) assisted evolutionary algorithms to generate domain-dependent generalized planners for classical planning tasks described in PDDL. By casting generalized planning as an optimization problem, GenePlan iteratively evolves interpretable Python planners that minimize plan length across diverse problem instances. In empirical evaluation across six existing benchmark domains and two new domains, GenePlan achieved an average SAT score of 0.91, closely matching the performance of the state-of-the-art planners (SAT score 0.93), and significantly outperforming other LLM-based baselines such as chain-of-thought (CoT) prompting (average SAT score 0.64). The generated planners solve new instances rapidly (average 0.49 seconds per task) and at low cost (average $1.82 per domain using GPT-4o). --- ### Prune Redundancy, Preserve Essence: Vision Token Compression in VLMs via Synergistic Importance-Diversity Authors: Zhengyao Fang, Pengyuan Lyu, Chengquan Zhang, Guangming Lu, Jun Yu, Wenjie Pei Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09480v2 Vision-language models (VLMs) face significant computational inefficiencies caused by excessive generation of visual tokens. While prior work shows that a large fraction of visual tokens are redundant, existing compression methods struggle to balance importance preservation and information diversity. To address this, we propose PruneSID, a training-free Synergistic Importance-Diversity approach featuring a two-stage pipeline: (1) Principal Semantic Components Analysis (PSCA) for clustering tokens into semantically coherent groups, ensuring comprehensive concept coverage, and (2) Intra-group Non-Maximum Suppression (NMS) for pruning redundant tokens while preserving key representative tokens within each group. Additionally, PruneSID incorporates an information-aware dynamic compression ratio mechanism that optimizes token compression rates based on image complexity, enabling more effective average information preservation across diverse scenes. Extensive experiments demonstrate state-of-the-art performance, achieving 96.3% accuracy on LLaVA-1.5 with only 11.1% token retention, and 92.8% accuracy at extreme compression rates (5.6%) on LLaVA-NeXT, outperforming prior methods by 2.5% with 7.8 $\times$ faster prefilling speed compared to the original model. Our framework generalizes across diverse VLMs and both image and video modalities, showcasing strong cross-modal versatility. Code is available at https://github.com/ZhengyaoFang/PruneSID. --- ### Telogenesis: Goal Is All U Need Authors: Zhuoran Deng, Yizhi Zhang, Ziyi Zhang, Wan Shen Published: 2026-03-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.09476v1 Goal-conditioned systems assume goals are provided externally. We ask whether attentional priorities can emerge endogenously from an agent's internal cognitive state. We propose a priority function that generates observation targets from three epistemic gaps: ignorance (posterior variance), surprise (prediction error), and staleness (temporal decay of confidence in unobserved variables). We validate this in two systems: a minimal attention-allocation environment (2,000 runs) and a modular, partially observable world (500 runs). Ablation shows each component is necessary. A key finding is metric-dependent reversal: under global prediction error, coverage-based rotation wins; under change detection latency, priority-guided allocation wins, with advantage growing monotonically with dimensionality (d = -0.95 at N=48, p < 10^-6). Detection latency follows a power law in attention budget, with a steeper exponent for priority-guided allocation (0.55 vs. 0.40). When the decay rate is made learnable per variable, the system spontaneously recovers environmental volatility structure without supervision (t = 22.5, p < 10^-6). We demonstrate that epistemic gaps alone, without external reward, suffice to generate adaptive priorities that outperform fixed strategies and recover latent environmental structure. --- ### OmniEarth: A Benchmark for Evaluating Vision-Language Models in Geospatial Tasks Authors: Ronghao Fu, Haoran Liu, Weijie Zhang, Zhiwen Lin, Xiao Yang, Peng Zhang, Bo Yang Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09471v1 Vision-Language Models (VLMs) have demonstrated effective perception and reasoning capabilities on general-domain tasks, leading to growing interest in their application to Earth observation. However, a systematic benchmark for comprehensively evaluating remote sensing vision-language models (RSVLMs) remains lacking. To address this gap, we introduce OmniEarth, a benchmark for evaluating RSVLMs under realistic Earth observation scenarios. OmniEarth organizes tasks along three capability dimensions: perception, reasoning, and robustness. It defines 28 fine-grained tasks covering multi-source sensing data and diverse geospatial contexts. The benchmark supports two task formulations: multiple-choice VQA and open-ended VQA. The latter includes pure text outputs for captioning tasks, bounding box outputs for visual grounding tasks, and mask outputs for segmentation tasks. To reduce linguistic bias and examine whether model predictions rely on visual evidence, OmniEarth adopts a blind test protocol and a quintuple semantic consistency requirement. OmniEarth includes 9,275 carefully quality-controlled images, including proprietary satellite imagery from Jilin-1 (JL-1), along with 44,210 manually verified instructions. We conduct a systematic evaluation of contrastive learning-based models, general closed-source and open-source VLMs, as well as RSVLMs. Results show that existing VLMs still struggle with geospatially complex tasks, revealing clear gaps that need to be addressed for remote sensing applications. OmniEarth is publicly available at https://huggingface.co/datasets/sjeeudd/OmniEarth. --- ### The Patrologia Graeca Corpus: OCR, Annotation, and Open Release of Noisy Nineteenth-Century Polytonic Greek Editions Authors: Chahan Vidal-Gorène, Bastien Kindt Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09470v1 We present the Patrologia Graeca Corpus, the first large-scale open OCR and linguistic resource for nineteenthcentury editions of Ancient Greek. The collection covers the remaining undigitized volumes of the Patrologia Graeca (PG), printed in complex bilingual (Greek-Latin) layouts and characterized by highly degraded polytonic Greek typography. Through a dedicated pipeline combining YOLO-based layout detection and CRNN-based text recognition, we achieve a character error rate (CER) of 1.05% and a word error rate (WER) of 4.69%, largely outperforming existing OCR systems for polytonic Greek. The resulting corpus contains around six million lemmatized and part-of-speech tagged tokens, aligned with full OCR and layout annotations. Beyond its philological value, this corpus establishes a new benchmark for OCR on noisy polytonic Greek and provides training material for future models, including LLMs. --- ### TopoOR: A Unified Topological Scene Representation for the Operating Room Authors: Tony Danjun Wang, Ka Young Kim, Tolga Birdal, Nassir Navab, Lennart Bastian Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09466v1 Surgical Scene Graphs abstract the complexity of surgical operating rooms (OR) into a structure of entities and their relations, but existing paradigms suffer from strictly dyadic structural limitations. Frameworks that predominantly rely on pairwise message passing or tokenized sequences flatten the manifold geometry inherent to relational structures and lose structure in the process. We introduce TopoOR, a new paradigm that models multimodal operating rooms as a higher-order structure, innately preserving pairwise and group relationships. By lifting interactions between entities into higher-order topological cells, TopoOR natively models complex dynamics and multimodality present in the OR. This topological representation subsumes traditional scene graphs, thereby offering strictly greater expressivity. We also propose a higher-order attention mechanism that explicitly preserves manifold structure and modality-specific features throughout hierarchical relational attention. In this way, we circumvent combining 3D geometry, audio, and robot kinematics into a single joint latent representation, preserving the precise multimodal structure required for safety-critical reasoning, unlike existing methods. Extensive experiments demonstrate that our approach outperforms traditional graph and LLM-based baselines across sterility breach detection, robot phase prediction, and next-action anticipation --- ### EvoDriveVLA: Evolving Driving VLA Models via Collaborative Perception-Planning Distillation Authors: Jiajun Cao, Xiaoan Zhang, Xiaobao Wei, Liyuqiu Huang, Zijian Wang, Hanzhen Zhang, Zhengyu Jia, Wei Mao, Hao Wang, Xianming Liu, Shuchang Zhou, Yang Wang et al. Published: 2026-03-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2603.09465v3 Vision-Language-Action models have shown great promise for autonomous driving, yet they suffer from degraded perception after unfreezing the visual encoder and struggle with accumulated instability in long-term planning. To address these challenges, we propose EvoDriveVLA-a novel collaborative perception-planning distillation framework that integrates self-anchored perceptual constraints and future-informed trajectory optimization. Specifically, self-anchored visual distillation leverages self-anchor teacher to deliver visual anchoring constraints, regularizing student representations via trajectory-guided key-region awareness. In parallel, future-informed trajectory distillation employs a future-aware oracle teacher with coarse-to-fine trajectory refinement and Monte Carlo dropout sampling to synthesize reasoning trajectories that model future evolutions, enabling the student model to internalize the future-aware insights of the teacher. EvoDriveVLA achieves SOTA performance in nuScenes open-loop evaluation and significantly enhances performance in NAVSIM closed-loop evaluation. Our code is available at: https://github.com/hey-cjj/EvoDriveVLA. --- ### An Empirical Study and Theoretical Explanation on Task-Level Model-Merging Collapse Authors: Yuan Cao, Dezhi Ran, Yuzhe Guo, Mengzhou Wu, Simin Chen, Linyi Li, Wei Yang, Tao Xie Published: 2026-03-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.09463v1 Model merging unifies independently fine-tuned LLMs from the same base, enabling reuse and integration of parallel development efforts without retraining. However, in practice we observe that merging does not always succeed: certain combinations of task-specialist models suffer from catastrophic performance degradation after merging. We refer to this failure mode as merging collapse. Intuitively, collapse arises when the learned representations or parameter adjustments for different tasks are fundamentally incompatible, so that merging forces destructive interference rather than synergy. In this paper, we identify and characterize the phenomenon of task-level merging collapse, where certain task combinations consistently trigger huge performance degradation across all merging methods. Through extensive experiments and statistical analysis, we demonstrate that representational incompatibility between tasks is strongly correlated with merging collapse, while parameter-space conflict metrics show minimal correlation, challenging conventional wisdom in model merging literature. We provide a theoretical explanation on this phenomenon through rate-distortion theory with a dimension-dependent bound, establishing fundamental limits on task mergeability regardless of methodology. --- ### Declarative Scenario-based Testing with RoadLogic Authors: Ezio Bartocci, Alessio Gambi, Felix Gigler, Cristinel Mateis, Dejan Ničković Published: 2026-03-10 Categories: cs.SE, cs.AI, cs.LO arXiv: https://arxiv.org/abs/2603.09455v2 Scenario-based testing is a key method for cost-effective and safe validation of autonomous vehicles (AVs). Existing approaches rely on imperative scenario definitions, requiring developers to manually enumerate numerous variants to achieve coverage. Declarative languages, such as ASAM OpenSCENARIO DSL (OS2), raise the abstraction level but lack systematic methods for instantiating concrete and specification-compliant scenarios. To our knowledge, currently, no open-source solution provides this capability. We present RoadLogic that bridges declarative OS2 specifications and executable simulations. It uses Answer Set Programming to generate abstract plans satisfying scenario constraints, motion planning to refine the plans into feasible trajectories, and specification-based monitoring to verify correctness. We evaluate RoadLogic on instantiating representative OS2 scenarios executed in the CommonRoad framework. Results show that RoadLogic consistently produces realistic, specification-satisfying simulations within minutes and captures diverse behavioral variants through parameter sampling, thus opening the door to systematic scenario-based testing for autonomous driving systems. --- ### Variational Routing: A Scalable Bayesian Framework for Calibrated Mixture-of-Experts Transformers Authors: Albus Yizhuo Li, Matthew Wicker Published: 2026-03-10 Categories: cs.LG, cs.AI, stat.ML arXiv: https://arxiv.org/abs/2603.09453v3 Foundation models are increasingly being deployed in contexts where understanding the uncertainty of their outputs is critical to ensuring responsible deployment. While Bayesian methods offer a principled approach to uncertainty quantification, their computational overhead renders their use impractical for training or inference at foundation model scale. State-of-the-art models achieve parameter counts in the trillions through carefully engineered sparsity including Mixture-of-Experts (MoE) layers. In this work, we demonstrate calibrated uncertainty at scale by introducing Variational Mixture-of-Experts Routing (VMoER), a structured Bayesian approach for modelling uncertainty in MoE layers. VMoER confines Bayesian inference to the expert-selection stage which is typically done by a deterministic routing network. We instantiate VMoER using two inference strategies: amortised variational inference over routing logits and inferring a temperature parameter for stochastic expert selection. Across fine-tuning tested foundation models, VMoER improves routing stability under noise by 38\%, reduces calibration error by 94\%, and increases out-of-distribution AUROC by 12\%, while incurring less than 1\% additional FLOPs. These results suggest VMoER offers a scalable path toward robust and uncertainty-aware foundation models. --- ### CyberThreat-Eval: Can Large Language Models Automate Real-World Threat Research? Authors: Xiangsen Chen, Xuan Feng, Shuo Chen, Matthieu Maitre, Sudipto Rakshit, Diana Duvieilh, Ashley Picone, Nan Tang Published: 2026-03-10 Categories: cs.CR, cs.CL arXiv: https://arxiv.org/abs/2603.09452v1 Analyzing Open Source Intelligence (OSINT) from large volumes of data is critical for drafting and publishing comprehensive CTI reports. This process usually follows a three-stage workflow -- triage, deep search and TI drafting. While Large Language Models (LLMs) offer a promising route toward automation, existing benchmarks still have limitations. These benchmarks often consist of tasks that do not reflect real-world analyst workflows. For example, human analysts rarely receive tasks in the form of multiple-choice questions. Also, existing benchmarks often rely on model-centric metrics that emphasize lexical overlap rather than actionable, detailed insights essential for security analysts. Moreover, they typically fail to cover the complete three-stage workflow. To address these issues, we introduce CyberThreat-Eval, which is collected from the daily CTI workflow of a world-leading company. This expert-annotated benchmark assesses LLMs on practical tasks across all three stages as mentioned above. It utilizes analyst-centric metrics that measure factual accuracy, content quality, and operational costs. Our evaluation using this benchmark reveals important insights into the limitations of current LLMs. For example, LLMs often lack the nuanced expertise required to handle complex details and struggle to distinguish between correct and incorrect information. To address these challenges, the CTI workflow incorporates both external ground-truth databases and human expert knowledge. TRA allows human experts to iteratively provide feedback for continuous improvement. The code is available at \href{https://github.com/xschen-beb/CyberThreat-Eval}{\texttt{GitHub}} and \href{https://huggingface.co/datasets/xse/CyberThreat-Eval}{\texttt{HuggingFace}}. --- ### A Guideline-Aware AI Agent for Zero-Shot Target Volume Auto-Delineation Authors: Yoon Jo Kim, Wonyoung Cho, Jongmin Lee, Han Joo Chae, Hyunki Park, Sang Hoon Seo, Jae Myung Noh, Kyungmi Yang, Dongryul Oh, Jin Sung Kim Published: 2026-03-10 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2603.09448v2 Delineating the clinical target volume (CTV) in radiotherapy involves complex margins constrained by tumor location and anatomical barriers. While deep learning models automate this process, their rigid reliance on expert-annotated data requires costly retraining whenever clinical guidelines update. To overcome this limitation, we introduce OncoAgent, a novel guideline-aware AI agent framework that seamlessly converts textual clinical guidelines into three-dimensional target contours in a training-free manner. Evaluated on esophageal cancer cases, the agent achieves a zero-shot Dice similarity coefficient of 0.842 for the CTV and 0.880 for the planning target volume, demonstrating performance highly comparable to a fully supervised nnU-Net baseline. Notably, in a blinded clinical evaluation, physicians strongly preferred OncoAgent over the supervised baseline, rating it higher in guideline compliance, modification effort, and clinical acceptability. Furthermore, the framework generalizes zero-shot to alternative esophageal guidelines and other anatomical sites (e.g., prostate) without any retraining. Beyond mere volumetric overlap, our agent-based paradigm offers near-instantaneous adaptability to alternative guidelines, providing a scalable and transparent pathway toward interpretability in radiotherapy treatment planning. --- ### GIIM: Graph-based Learning of Inter- and Intra-view Dependencies for Multi-view Medical Image Diagnosis Authors: Tran Bao Sam, Hung Vu, Dao Trung Kien, Tran Dat Dang, Van Ha Tang, Steven Truong Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09446v1 Computer-aided diagnosis (CADx) has become vital in medical imaging, but automated systems often struggle to replicate the nuanced process of clinical interpretation. Expert diagnosis requires a comprehensive analysis of how abnormalities relate to each other across various views and time points, but current multi-view CADx methods frequently overlook these complex dependencies. Specifically, they fail to model the crucial relationships within a single view and the dynamic changes lesions exhibit across different views. This limitation, combined with the common challenge of incomplete data, greatly reduces their predictive reliability. To address these gaps, we reframe the diagnostic task as one of relationship modeling and propose GIIM, a novel graph-based approach. Our framework is uniquely designed to simultaneously capture both critical intra-view dependencies between abnormalities and inter-view dynamics. Furthermore, it ensures diagnostic robustness by incorporating specific techniques to effectively handle missing data, a common clinical issue. We demonstrate the generality of this approach through extensive evaluations on diverse imaging modalities, including CT, MRI, and mammography. The results confirm that our GIIM model significantly enhances diagnostic accuracy and robustness over existing methods, establishing a more effective framework for future CADx systems. --- ### From Weighting to Modeling: A Nonparametric Estimator for Off-Policy Evaluation Authors: Rong J. B. Zhu Published: 2026-03-10 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.09436v1 We study off-policy evaluation in the setting of contextual bandits, where we aim to evaluate a new policy using historical data that consists of contexts, actions and received rewards. This historical data typically does not faithfully represent action distribution of the new policy accurately. A common approach, inverse probability weighting (IPW), adjusts for these discrepancies in action distributions. However, this method often suffers from high variance due to the probability being in the denominator. The doubly robust (DR) estimator reduces variance through modeling reward but does not directly address variance from IPW. In this work, we address the limitation of IPW by proposing a Nonparametric Weighting (NW) approach that constructs weights using a nonparametric model. Our NW approach achieves low bias like IPW but typically exhibits significantly lower variance. To further reduce variance, we incorporate reward predictions -- similar to the DR technique -- resulting in the Model-assisted Nonparametric Weighting (MNW) approach. The MNW approach yields accurate value estimates by explicitly modeling and mitigating bias from reward modeling, without aiming to guarantee the standard doubly robust property. Extensive empirical comparisons show that our approaches consistently outperform existing techniques, achieving lower variance in value estimation while maintaining low bias. --- ### AI Act Evaluation Benchmark: An Open, Transparent, and Reproducible Evaluation Dataset for NLP and RAG Systems Authors: Athanasios Davvetas, Michael Papademas, Xenia Ziouvelou, Vangelis Karkaletsis Published: 2026-03-10 Categories: cs.AI arXiv: https://arxiv.org/abs/2603.09435v1 The rapid rollout of AI in heterogeneous public and societal sectors has subsequently escalated the need for compliance with regulatory standards and frameworks. The EU AI Act has emerged as a landmark in the regulatory landscape. The development of solutions that elicit the level of AI systems' compliance with such standards is often limited by the lack of resources, hindering the semi-automated or automated evaluation of their performance. This generates the need for manual work, which is often error-prone, resource-limited or limited to cases not clearly described by the regulation. This paper presents an open, transparent, and reproducible method of creating a resource that facilitates the evaluation of NLP models with a strong focus on RAG systems. We have developed a dataset that contain the tasks of risk-level classification, article retrieval, obligation generation, and question-answering for the EU AI Act. The dataset files are in a machine-to-machine appropriate format. To generate the files, we utilise domain knowledge as an exegetical basis, combining with the processing and reasoning power of large language models to generate scenarios along with the respective tasks. Our methodology demonstrates a way to harness language models for grounded generation with high document relevancy. Besides, we overcome limitations such as navigating the decision boundaries of risk-levels that are not explicitly defined within the EU AI Act, such as limited and minimal cases. Finally, we demonstrate our dataset's effectiveness by evaluating a RAG-based solution that reaches 0.87 and 0.85 F1-score for prohibited and high-risk scenarios. --- ### Common Sense vs. Morality: The Curious Case of Narrative Focus Bias in LLMs Authors: Saugata Purkayastha, Pranav Kushare, Pragya Paramita Pal, Sukannya Purkayastha Published: 2026-03-10 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2603.09434v1 Large Language Models (LLMs) are increasingly deployed across diverse real-world applications and user communities. As such, it is crucial that these models remain both morally grounded and knowledge-aware. In this work, we uncover a critical limitation of current LLMs -- their tendency to prioritize moral reasoning over commonsense understanding. To investigate this phenomenon, we introduce CoMoral, a novel benchmark dataset containing commonsense contradictions embedded within moral dilemmas. Through extensive evaluation of ten LLMs across different model sizes, we find that existing models consistently struggle to identify such contradictions without prior signal. Furthermore, we observe a pervasive narrative focus bias, wherein LLMs more readily detect commonsense contradictions when they are attributed to a secondary character rather than the primary (narrator) character. Our comprehensive analysis underscores the need for enhanced reasoning-aware training to improve the commonsense robustness of large language models. --- ### Impact of Markov Decision Process Design on Sim-to-Real Reinforcement Learning Authors: Tatjana Krau, Jorge Mandlmaier, Tobias Damm, Frieder Heieck Published: 2026-03-10 Categories: cs.LG arXiv: https://arxiv.org/abs/2603.09427v2 Reinforcement Learning (RL) has demonstrated strong potential for industrial process control, yet policies trained in simulation often suffer from a significant sim-to-real gap when deployed on physical hardware. This work systematically analyzes how core Markov Decision Process (MDP) design choices -- state composition, target inclusion, reward formulation, termination criteria, and environment dynamics models -- affect this transfer. Using a color mixing task, we evaluate different MDP configurations and mixing dynamics across simulation and real-world experiments. We validate our findings on physical hardware, demonstrating that physics-based dynamics models achieve up to 50% real-world success under strict precision constraints where simplified models fail entirely. Our results provide practical MDP design guidelines for deploying RL in industrial process control. --- ### CERES: A Probabilistic Early Warning System for Acute Food Insecurity Authors: Tom Danny S. Pedersen Published: 2026-03-10 Categories: stat.AP, cs.AI arXiv: https://arxiv.org/abs/2603.09425v1 We present CERES (Calibrated Early-warning and Risk Estimation System), an automated probabilistic forecasting system for acute food insecurity. CERES generates 90-day ahead probability estimates of IPC Phase 3+ (Crisis), Phase 4+ (Emergency), and Phase 5 (Famine) conditions for 43 high-risk countries globally, updated weekly. The system fuses six data streams, precipitation anomalies (CHIRPS), vegetation indices (MODIS NDVI), conflict events (ACLED), IPC classifications, food consumption scores (WFP), and cereal price indices (FAO/WFP) - through a logistic scoring model with author-specified initial coefficients and parametric input-perturbation intervals (n=2,000 draws). In historical back-validation against four IPC Phase 4-5 events selected for data completeness, CERES assigned TIER-1 classification in all four cases; these are in-sample sanity checks only, not prospective performance claims. All prospective predictions are timestamped, cryptographically identified, and archived for public verification against IPC outcome data at the T+90 horizon. To the author's knowledge, CERES is the first famine early warning system that is simultaneously: (1) probabilistic, (2) open-access, (3) continuously running, (4) machine-readable at prediction level, and (5) committed to public prospective verification of every prediction made. --- ### Class-Incremental Motion Forecasting Authors: Nicolas Schischka, Nikhil Gosala, B Ravi Kiran, Senthil Yogamani, Abhinav Valada Published: 2026-03-10 Categories: cs.CV, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2603.09420v3 Motion forecasting enables autonomous vehicles to anticipate scene evolution by predicting the future trajectories of dynamic agents. However, existing approaches typically assume a closed-world setting with a fixed object taxonomy and access to high-quality perception, limiting their applicability in the real world where perception is imperfect, and new object classes may emerge over time. In this work, we introduce class-incremental motion forecasting, a novel setting in which new object classes are sequentially introduced over time and future object trajectories are predicted directly from camera images. We propose the first end-to-end framework for this setting, which adapts to newly introduced classes while mitigating catastrophic forgetting of previously learned ones. Our method generates motion forecasting pseudo-labels for known classes and matches them with 2D instance masks from an open-vocabulary segmentation model. This 3D-to-2D keypoint voting mechanism filters inconsistent and overconfident predictions, while a query feature variance-based replay strategy samples informative past sequences to preserve prior knowledge. Extensive evaluations on nuScenes and Argoverse 2 show that our approach successfully preserves performance on known classes while effectively adapting to novel ones. We further demonstrate zero-shot transfer to real-world driving and show that the framework extends naturally to open- and closed-loop end-to-end class-incremental planning on nuScenes and NeuroNCAP. Code and models will be made publicly available at https://omen.cs.uni-freiburg.de. --- ### MetaDAT: Generalizable Trajectory Prediction via Meta Pre-training and Data-Adaptive Test-Time Updating Authors: Yuning Wang, Pu Zhang, Yuan He, Ke Wang, Jianru Xue Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09419v1 Existing trajectory prediction methods exhibit significant performance degradation under distribution shifts during test time. Although test-time training techniques have been explored to enable adaptation, current approaches rely on an offline pre-trained predictor that lacks online learning flexibility. Moreover, they depend on fixed online model updating rules that do not accommodate the specific characteristics of test data. To address these limitations, we first propose a meta-learning framework to directly optimize the predictor for fast and accurate online adaptation, which performs bi-level optimization on the performance of simulated test-time adaptation tasks during pre-training. Furthermore, at test time, we introduce a data-adaptive model updating mechanism that dynamically adjusts the predefined learning rates and updating frequencies based on online partial derivatives and hard sample selection. This mechanism enables the online learning rate to suit the test data, and focuses on informative hard samples to enhance efficiency. Experiments are conducted on various challenging cross-dataset distribution shift scenarios, including nuScenes, Lyft, and Waymo. Results demonstrate that our method achieves superior adaptation accuracy, surpassing state-of-the-art test-time training methods for trajectory prediction. Additionally, our method excels under suboptimal learning rates and high FPS demands, showcasing its robustness and practicality. --- ### CIGPose: Causal Intervention Graph Neural Network for Whole-Body Pose Estimation Authors: Bohao Li, Zhicheng Cao, Huixian Li, Yangming Guo Published: 2026-03-10 Categories: cs.CV arXiv: https://arxiv.org/abs/2603.09418v2 State-of-the-art whole-body pose estimators often lack robustness, producing anatomically implausible predictions in challenging scenes. We posit this failure stems from spurious correlations learned from visual context, a problem we formalize using a Structural Causal Model (SCM). The SCM identifies visual context as a confounder that creates a non-causal backdoor path, corrupting the model's reasoning. We introduce the Causal Intervention Graph Pose (CIGPose) framework to address this by approximating the true causal effect between visual evidence and pose. The core of CIGPose is a novel Causal Intervention Module: it first identifies confounded keypoint representations via predictive uncertainty and then replaces them with learned, context-invariant canonical embeddings. These deconfounded embeddings are processed by a hierarchical graph neural network that reasons over the human skeleton at both local and global semantic levels to enforce anatomical plausibility. Extensive experiments show CIGPose achieves a new state-of-the-art on COCO-WholeBody. Notably, our CIGPose-x model achieves 67.0\% AP, surpassing prior methods that rely on extra training data. With the additional UBody dataset, CIGPose-x is further boosted to 67.5\% AP, demonstrating superior robustness and data efficiency. The codes and models are publicly available at https://github.com/53mins/CIGPose. --- ### Investigating Gender Stereotypes in Large Language Models via Social Determinants of Health Authors: Trung Hieu Ngo, Adrien Bazoge, Solen Quiniou, Pierre-Antoine Gourraud, Emmanuel Morin Published: 2026-03-10 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2603.09416v1 Large Language Models (LLMs) excel in Natural Language Processing (NLP) tasks, but they often propagate biases embedded in their training data, which is potentially impactful in sensitive domains like healthcare. While existing benchmarks evaluate biases related to individual social determinants of health (SDoH) such as gender or ethnicity, they often overlook interactions between these factors and lack context-specific assessments. This study investigates bias in LLMs by probing the relationships between gender and other SDoH in French patient records. Through a series of experiments, we found that embedded stereotypes can be probed using SDoH input and that LLMs rely on embedded stereotypes to make gendered decisions, suggesting that evaluating interactions among SDoH factors could usefully complement existing approaches to assessing LLM performance and bias. ---