# Research Papers — 2026-04 · bundle page 186 of 233

> Full-text markdown of papers #9251–9300 of 11612 published in 2026-04, newest first. AI-readable mirrors of the full 11612-paper 2026-04 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-04/page/186
[← Previous bundle](https://aipod.dev/papers/archive/2026-04/page/185.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-04/page/187.md)

### What Models Know, How Well They Know It: Knowledge-Weighted Fine-Tuning for Learning When to Say "I Don't Know"

Authors: Joosung Lee, Hwiyeol Jo, Donghyeon Ko, Kyubyung Chae, Cheonbok Park, Jeonghoon Kim
Published: 2026-04-07
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2604.05779v2

While large language models (LLMs) demonstrate strong capabilities across diverse user queries, they still suffer from hallucinations, often arising from knowledge misalignment between pre-training and fine-tuning. To address this misalignment, we reliably estimate a fine-grained, instance-level knowledge score via multi-sampled inference. Using the knowledge score, we scale the learning signal according to the model's existing knowledge, while encouraging explicit "I don't know" responses for out-of-scope queries. Experimental results show that this approach allows the model to explicitly express uncertainty when it lacks knowledge, while maintaining accuracy on questions it can answer. Furthermore, we propose evaluation metrics for uncertainty, showing that accurate discrimination between known and unknown instances consistently improves performance.

---

### What Models Know, How Well They Know It: Knowledge-Weighted Fine-Tuning for Learning When to Say "I Don't Know"

Authors: Joosung Lee, Hwiyeol Jo, Donghyeon Ko, Kyubyung Chae, Cheonbok Park, Jeonghoon Kim
Published: 2026-04-07
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2604.05779v1

While large language models (LLMs) demonstrate strong capabilities across diverse user queries, they still suffer from hallucinations, often arising from knowledge misalignment between pre-training and fine-tuning. To address this misalignment, we reliably estimate a fine-grained, instance-level knowledge score via multi-sampled inference. Using the knowledge score, we scale the learning signal according to the model's existing knowledge, while encouraging explicit "I don't know" responses for out-of-scope queries. Experimental results show that this approach allows the model to explicitly express uncertainty when it lacks knowledge, while maintaining accuracy on questions it can answer. Furthermore, we propose evaluation metrics for uncertainty, showing that accurate discrimination between known and unknown instances consistently improves performance.

---

### Effective Dynamics and Transition Pathways from Koopman-Inspired Neural Learning of Collective Variables

Authors: Alexander Sikorski, Luca Donati, Marcus Weber, Christof Schütte
Published: 2026-04-07
Categories: math.DS, physics.chem-ph, stat.ML
arXiv: https://arxiv.org/abs/2604.05778v1

The ISOKANN (Invariant Subspaces of Koopman Operators Learned by Artificial Neural Networks) framework provides a data-driven route to extract collective variables (CVs) and effective dynamics from complex molecular systems. In this work, we integrate the theoretical foundation of Koopman operators with Krylov-like subspace algorithms, and reduced dynamical modeling to build a coherent picture of how to describe metastable transitions in high-dimensional systems based on CVs. Starting from the identification of CVs based on dominant invariant subspaces, we derive the corresponding effective dynamics on the latent space and connect these to transition rates and times, committor functions, and transition pathways. The combination of Koopman-based learning and reduced-dimensional effective dynamics yields a principled framework for computing transition rates and pathways from simulation data. Numerical experiments on one-, two-, and three-dimensional benchmark potentials illustrate the ability of ISOKANN to reconstruct the coarse-grained kinetics and reproduce transition times across enthalpic and entropic barriers.

---

### Emergent social transmission of model-based representations without inference

Authors: Silja Keßler, Miriam Bautista-Salinero, Claudio Tennie, Charley M. Wu
Published: 2026-04-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.05777v2

How do people acquire rich, flexible knowledge about their environment from others despite limited cognitive capacity? Humans are often thought to rely on computationally costly mentalizing, such as inferring others' beliefs. In contrast, cultural evolution emphasizes that behavioral transmission can be supported by simple social cues. Using reinforcement learning simulations, we show how minimal social learning can indirectly transmit higher-level representations. We simulate a naïve agent searching for rewards in a reconfigurable environment, learning either alone or by observing an expert - crucially, without inferring mental states. Instead, the learner heuristically selects actions or boosts value representations based on observed actions. Our results demonstrate that these cues bias the learner's experience, causing its representation to converge toward the expert's. Model-based learners benefit most from social exposure, showing faster learning and more expert-like representations. These findings show how cultural transmission can arise from simple, non-mentalizing processes exploiting asocial learning mechanisms.

---

### PhageBench: Can LLMs Understand Raw Bacteriophage Genomes?

Authors: Yusen Hou, Weicai Long, Haitao Hu, Houcheng Su, Junning Feng, Yanlin Zhang
Published: 2026-04-07
Categories: cs.CL, q-bio.GN
arXiv: https://arxiv.org/abs/2604.05775v1

Bacteriophages, often referred to as the dark matter of the biosphere, play a critical role in regulating microbial ecosystems and in antibiotic alternatives. Thus, accurate interpretation of their genomes holds significant scientific and practical value. While general-purpose Large Language Models (LLMs) excel at understanding biological texts, their ability to directly interpret raw nucleotide sequences and perform biological reasoning remains underexplored. To address this, we introduce PhageBench, the first benchmark designed to evaluate phage genome understanding by mirroring the workflow of bioinformatics experts. The dataset contains 5,600 high-quality samples covering five core tasks across three stages: Screening, Quality Control, and Phenotype Annotation. Our evaluation of eight LLMs reveals that general-purpose reasoning models significantly outperform random baselines in phage contig identification and host prediction, demonstrating promising potential for genomic understanding. However, they exhibit significant limitations in complex reasoning tasks involving long-range dependencies and fine-grained functional localization. These findings highlight the necessity of developing next-generation models with enhanced reasoning capabilities for biological sequences.

---

### GenomeQA: Benchmarking General Large Language Models for Genome Sequence Understanding

Authors: Weicai Long, Yusen Hou, Junning Feng, Houcheng Su, Shuo Yang, Donglin Xie, Yanlin Zhang
Published: 2026-04-07
Categories: q-bio.GN, cs.CL
arXiv: https://arxiv.org/abs/2604.05774v1

Large Language Models (LLMs) are increasingly adopted as conversational assistants in genomics, where they are mainly used to reason over biological knowledge, annotations, and analysis outputs through natural language interfaces. However, existing benchmarks either focus on specialized DNA models trained for sequence prediction or evaluate biological knowledge using text-only questions, leaving the behavior of general-purpose LLMs when directly exposed to raw genome sequences underexplored. We introduce GenomeQA, a benchmark designed to provide a controlled evaluation setting for general-purpose LLMs on sequence-based genome inference tasks. GenomeQA comprises 5,200 samples drawn from multiple biological databases, with sequence lengths ranging from 6 to 1,000 base pairs (bp), spanning six task families: Enhancer and Promoter Identification, Splice Site Identification, Taxonomic Classification, Histone Mark Prediction, Transcription Factor Binding Site Prediction, and TF Motif Prediction. Across six frontier LLMs, we find that models consistently outperform random baselines and can exploit local sequence signals such as GC content and short motifs, while performance degrades on tasks that require more indirect or multi-step inference over sequence patterns. GenomeQA establishes a diagnostic benchmark for studying and improving the use of general-purpose LLMs on raw genomic sequences.

---

### PDMP: Rethinking Balanced Multimodal Learning via Performance-Dominant Modality Prioritization

Authors: Shicai Wei, Chunbo Luo, Qiang Zhu, Yang Luo
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05773v1

Multimodal learning has attracted increasing attention due to its practicality. However, it often suffers from insufficient optimization, where the multimodal model underperforms even compared to its unimodal counterparts. Existing methods attribute this problem to the imbalanced learning between modalities and solve it by gradient modulation. This paper argues that balanced learning is not the optimal setting for multimodal learning. On the contrary, imbalanced learning driven by the performance-dominant modality that has superior unimodal performance can contribute to better multimodal performance. And the under-optimization problem is caused by insufficient learning of the performance-dominant modality. To this end, we propose the Performance-Dominant Modality Prioritization (PDMP) strategy to assist multimodal learning. Specifically, PDMP firstly mines the performance-dominant modality via the performance ranking of the independently trained unimodal model. Then PDMP introduces asymmetric coefficients to modulate the gradients of each modality, enabling the performance-dominant modality to dominate the optimization. Since PDMP only relies on the unimodal performance ranking, it is independent of the structures and fusion methods of the multimodal model and has great potential for practical scenarios. Finally, extensive experiments on various datasets validate the superiority of PDMP.

---

### Beyond the Beep: Scalable Collision Anticipation and Real-Time Explainability with BADAS-2.0

Authors: Roni Goldshmidt, Hamish Scott, Lorenzo Niccolini, Hernan Matzner
Published: 2026-04-07
Categories: cs.CV, cs.CL
arXiv: https://arxiv.org/abs/2604.05767v2

We present BADAS-2.0, the second generation of our collision anticipation system, building on BADAS-1.0, which showed that fine-tuning V-JEPA2 on large-scale ego-centric dashcam data outperforms both academic baselines and production ADAS systems. BADAS-2.0 advances the state of the art along three axes. (i) Long-tail benchmark and accuracy: We introduce a 10-group long-tail benchmark targeting rare and safety-critical scenarios. To construct it, BADAS-1.0 is used as an active oracle to score millions of unlabeled drives and surface high-risk candidates for annotation. Combined with Nexar's Atlas platform for targeted data collection, this expands the dataset from 40k to 178,500 labeled videos (~2M clips), yielding consistent gains across all subgroups, with the largest improvements on the hardest long-tail cases. (ii) Knowledge distillation to edge: Domain-specific self-supervised pre-training on 2.25M unlabeled driving videos enables distillation into compact models, BADAS-2.0-Flash (86M) and BADAS-2.0-Flash-Lite (22M), achieving 7-12x speedup with near-parity accuracy, enabling real-time edge deployment. (iii) Explainability: BADAS-2.0 produces real-time object-centric attention heatmaps that localize the evidence behind predictions. BADAS-Reason extends this with a vision-language model that consumes the last frame and heatmap to generate driver actions and structured textual reasoning. Inference code and evaluation benchmarks are publicly available.

---

### Improving Controllable Generation: Faster Training and Better Performance via $x_0$-Supervision

Authors: Amadou S. Sangare, Adrien Maglo, Mohamed Chaouch, Bertrand Luvison
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05761v1

Text-to-Image (T2I) diffusion/flow models have recently achieved remarkable progress in visual fidelity and text alignment. However, they remain limited when users need to precisely control image layouts, something that natural language alone cannot reliably express. Controllable generation methods augment the initial T2I model with additional conditions that more easily describe the scene. Prior works straightforwardly train the augmented network with the same loss as the initial network. Although natural at first glance, this can lead to very long training times in some cases before convergence. In this work, we revisit the training objective of controllable diffusion models through a detailed analysis of their denoising dynamics. We show that direct supervision on the clean target image, dubbed $x_0$-supervision, or an equivalent re-weighting of the diffusion loss, yields faster convergence. Experiments on multiple control settings demonstrate that our formulation accelerates convergence by up to 2$\times$ according to our novel metric (mean Area Under the Convergence Curve - mAUCC), while also improving both visual quality and conditioning accuracy. Our code is available at https://github.com/CEA-LIST/x0-supervision

---

### High-dimensional reliability-based design optimization using stochastic emulators

Authors: M. Moustapha, B. Sudret
Published: 2026-04-07
Categories: stat.CO, stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2604.05759v1

Reliability-based design optimization (RBDO) is traditionally formulated as a nested optimization and reliability problem. Although surrogate models are generally employed to improve efficiency, the approach remains computationally prohibitive in high-dimensional settings. This paper proposes a novel RBDO framework based on a stochastic simulator viewpoint, in which the deterministic limit-state function and the uncertainty in the model inputs are combined into a unified stochastic representation. Under this formulation, the system response conditioned on a given design is modeled directly through its output distribution, rather than through an explicit limit-state function. Stochastic emulators are constructed in the design space to approximate the conditional response distribution, enabling the semi-analytical evaluation of failure probabilities or associated quantiles without resorting to Monte Carlo simulation. Two classes of stochastic emulators are investigated, namely generalized lambda models and stochastic polynomial chaos expansions. Both approaches provide a deterministic mapping between design variables and reliability constraints, which breaks the classical double-loop structure of RBDO and allows the use of standard deterministic optimization algorithms. The performance of the proposed approach is evaluated on a set of benchmark problems with dimensionality ranging from low to very high, including a case with stochastic excitation. The results are compared against a Kriging-based approach formulated in the full input space. The proposed method yields substantial computational gains, particularly in high-dimensional settings. While its efficiency is comparable to Kriging for low-dimensional problems, it significantly outperforms Kriging as the dimensionality increases.

---

### Identifying Influential N-grams in Confidence Calibration via Regression Analysis

Authors: Shintaro Ozaki, Wataru Hashimoto, Hidetaka Kamigaito, Katsuhiko Hayashi, Taro Watanabe
Published: 2026-04-07
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.05757v1

While large language models (LLMs) improve performance by explicit reasoning, their responses are often overconfident, even though they include linguistic expressions demonstrating uncertainty. In this work, we identify what linguistic expressions are related to confidence by applying the regression method. Specifically, we predict confidence of those linguistic expressions in the reasoning parts of LLMs as the dependent variables and analyze the relationship between a specific $n$-gram and confidence. Across multiple models and QA benchmarks, we show that LLMs remain overconfident when reasoning is involved and attribute this behavior to specific linguistic information. Interestingly, several of the extracted expressions coincide with cue phrases intentionally inserted on test-time scaling to improve reasoning performance. Through our test on causality and verification that the extracted linguistic information truly affects confidence, we reveal that confidence calibration is possible by simply suppressing those overconfident expressions without drops in performance.

---

### Controlling Distributional Bias in Multi-Round LLM Generation via KL-Optimized Fine-Tuning

Authors: Yanbei Jiang, Amr Keleg, Ryandito Diandaru, Jey Han Lau, Lea Frermann, Biaoyan Fang, Fajri Koto
Published: 2026-04-07
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.05756v1

While the real world is inherently stochastic, Large Language Models (LLMs) are predominantly evaluated on single-round inference against fixed ground truths. In this work, we shift the lens to distribution alignment: assessing whether LLMs, when prompted repeatedly, can generate outputs that adhere to a desired target distribution, e.g. reflecting real-world statistics or a uniform distribution. We formulate distribution alignment using the attributes of gender, race, and sentiment within occupational contexts. Our empirical analysis reveals that off-the-shelf LLMs and standard alignment techniques, including prompt engineering and Direct Preference Optimization, fail to reliably control output distributions. To bridge this gap, we propose a novel fine-tuning framework that couples Steering Token Calibration with Semantic Alignment. We introduce a hybrid objective function combining Kullback-Leibler divergence to anchor the probability mass of latent steering tokens and Kahneman-Tversky Optimization to bind these tokens to semantically consistent responses. Experiments across six diverse datasets demonstrate that our approach significantly outperforms baselines, achieving precise distributional control in attribute generation tasks.

---

### CAKE: Cloud Architecture Knowledge Evaluation of Large Language Models

Authors: Tim Lukas Adam, Phongsakon Mark Konrad, Riccardo Terrenzi, Florian Girardo Lukas, Rahime Yilmaz, Krzysztof Sierszecki, Serkan Ayvaz
Published: 2026-04-07
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2604.05755v1

In today's software architecture, large language models (LLMs) serve as software architecture co-pilots. However, no benchmark currently exists to evaluate large language models' actual understanding of cloud-native software architecture. For this reason we present a benchmark called CAKE, which consists of 188 expert-validated questions covering four cognitive levels of Bloom's revised taxonomy -- recall, analyze, design, and implement -- and five cloud-native topics. Evaluation is conducted on 22 model configurations (0.5B--70B parameters) across four LLM families, using three-run majority voting for multiple-choice questions (MCQs) and LLM-as-a-judge scoring for free-responses (FR). Based on this evaluation, four notable findings were identified. First, MCQ accuracy plateaus above 3B parameters, with the best model reaching 99.2\%. Second, free-response scores scale steadily across all cognitive levels. Third, the two formats capture different facets of knowledge, as the MCQ accuracy approaches a ceiling while free-responses continue to differentiate models. Finally, reasoning augmentation (+think) improves free-response quality, while tool augmentation (+tool) degrades performance for small models. These results suggest that the evaluation format fundamentally shapes how we measure architectural knowledge in LLMs.

---

### Brain-to-Speech: Prosody Feature Engineering and Transformer-Based Reconstruction

Authors: Mohammed Salah Al-Radhi, Géza Németh, Andon Tchechmedjiev, Binbin Xu
Published: 2026-04-07
Categories: eess.SP, cs.LG, cs.SD
arXiv: https://arxiv.org/abs/2604.05751v1

This chapter presents a novel approach to brain-to-speech (BTS) synthesis from intracranial electroencephalography (iEEG) data, emphasizing prosody-aware feature engineering and advanced transformer-based models for high-fidelity speech reconstruction. Driven by the increasing interest in decoding speech directly from brain activity, this work integrates neuroscience, artificial intelligence, and signal processing to generate accurate and natural speech. We introduce a novel pipeline for extracting key prosodic features directly from complex brain iEEG signals, including intonation, pitch, and rhythm. To effectively utilize these crucial features for natural-sounding speech, we employ advanced deep learning models. Furthermore, this chapter introduces a novel transformer encoder architecture specifically designed for brain-to-speech tasks. Unlike conventional models, our architecture integrates the extracted prosodic features to significantly enhance speech reconstruction, resulting in generated speech with improved intelligibility and expressiveness. A detailed evaluation demonstrates superior performance over established baseline methods, such as traditional Griffin-Lim and CNN-based reconstruction, across both quantitative and perceptual metrics. By demonstrating these advancements in feature extraction and transformer-based learning, this chapter contributes to the growing field of AI-driven neuroprosthetics, paving the way for assistive technologies that restore communication for individuals with speech impairments. Finally, we discuss promising future research directions, including the integration of diffusion models and real-time inference systems.

---

### SVC 2026: the Second Multimodal Deception Detection Challenge and the First Domain Generalized Remote Physiological Measurement Challenge

Authors: Dongliang Zhu, Zhiyi Niu, Bo Zhao, Jiajian Huang, Shuo Ye, Xun Lin, Hui Ma, Taorui Wang, Jiayu Zhang, Chunmei Zhu, Junzhe Cao, Yingjie Ma et al.
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05748v1

Subtle visual signals, although difficult to perceive with the naked eye, contain important information that can reveal hidden patterns in visual data. These signals play a key role in many applications, including biometric security, multimedia forensics, medical diagnosis, industrial inspection, and affective computing. With the rapid development of computer vision and representation learning techniques, detecting and interpreting such subtle signals has become an emerging research direction. However, existing studies often focus on specific tasks or modalities, and models still face challenges in robustness, representation ability, and generalization when handling subtle and weak signals in real-world environments. To promote research in this area, we organize the Subtle visual Challenge, which aims to learn robust representations for subtle visual signals. The challenge includes two tasks: cross-domain multimodal deception detection and remote photoplethysmography (rPPG) estimation. We hope that this challenge will encourage the development of more robust and generalizable models for subtle visual understanding, and further advance research in computer vision and multimodal learning. A total of 22 teams submitted their final results to this workshop competition, and the corresponding baseline models have been released on the \href{https://sites.google.com/view/svc-cvpr26}{MMDD2026 platform}\footnote{https://sites.google.com/view/svc-cvpr26}

---

### On the Robustness of Diffusion-Based Image Compression to Bit-Flip Errors

Authors: Amit Vaisman, Gal Pomerants, Raz Lapid
Published: 2026-04-07
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2604.05743v2

Modern image compression methods are typically optimized for the rate--distortion--perception trade-off, whereas their robustness to bit-level corruption is rarely examined. We show that diffusion-based compressors built on the Reverse Channel Coding (RCC) paradigm are substantially more robust to bit flips than classical and learned codecs. We further introduce a more robust variant of Turbo-DDCM that significantly improves robustness while only minimally affecting the rate--distortion--perception trade-off. Our findings suggest that RCC-based compression can yield more resilient compressed representations, potentially reducing reliance on error-correcting codes in highly noisy environments.

---

### ASSR-Net: Anisotropic Structure-Aware and Spectrally Recalibrated Network for Hyperspectral Image Fusion

Authors: Qiya Song, Hongzhi Zhou, Lishan Tan, Renwei Dian, Shutao Li
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05742v1

Hyperspectral image fusion aims to reconstruct high-spatial-resolution hyperspectral images (HR-HSI) by integrating complementary information from multi-source inputs. Despite recent progress, existing methods still face two critical challenges: (1) inadequate reconstruction of anisotropic spatial structures, resulting in blurred details and compromised spatial quality; and (2) spectral distortion during fusion, which hinders fine-grained spectral representation. To address these issues, we propose \textbf{ASSR-Net}: an Anisotropic Structure-Aware and Spectrally Recalibrated Network for Hyperspectral Image Fusion. ASSR-Net adopts a two-stage fusion strategy comprising anisotropic structure-aware spatial enhancement (ASSE) and hierarchical prior-guided spectral calibration (HPSC). In the first stage, a directional perception fusion module adaptively captures structural features along multiple orientations, effectively reconstructing anisotropic spatial patterns. In the second stage, a spectral recalibration module leverages the original low-resolution HSI as a spectral prior to explicitly correct spectral deviations in the fused results, thereby enhancing spectral fidelity. Extensive experiments on various benchmark datasets demonstrate that ASSR-Net consistently outperforms state-of-the-art methods, achieving superior spatial detail preservation and spectral consistency.

---

### MedLayBench-V: A Large-Scale Benchmark for Expert-Lay Semantic Alignment in Medical Vision Language Models

Authors: Han Jang, Junhyeok Lee, Heeseong Eum, Kyu Sung Choi
Published: 2026-04-07
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.05738v2

Medical Vision-Language Models (Med-VLMs) have achieved expert-level proficiency in interpreting diagnostic imaging. However, current models are predominantly trained on professional literature, limiting their ability to communicate findings in the lay register required for patient-centered care. While text-centric research has actively developed resources for simplifying medical jargon, there is a critical absence of large-scale multimodal benchmarks designed to facilitate lay-accessible medical image understanding. To bridge this resource gap, we introduce MedLayBench-V, the first large-scale multimodal benchmark dedicated to expert-lay semantic alignment. Unlike naive simplification approaches that risk hallucination, our dataset is constructed via a Structured Concept-Grounded Refinement (SCGR) pipeline. This method enforces strict semantic equivalence by integrating Unified Medical Language System (UMLS) Concept Unique Identifiers (CUIs) with micro-level entity constraints. MedLayBench-V provides a verified foundation for training and evaluating next-generation Med-VLMs capable of bridging the communication divide between clinical experts and patients.

---

### Graph Topology Information Enhanced Heterogeneous Graph Representation Learning

Authors: He Zhao, Zhiwei Zeng, Yongwei Wang, Chunyan Miao
Published: 2026-04-07
Categories: cs.LG, cs.IR
arXiv: https://arxiv.org/abs/2604.05732v1

Real-world heterogeneous graphs are inherently noisy and usually not in the optimal graph structures for downstream tasks, which often adversely affects the performance of GRL models in downstream tasks. Although Graph Structure Learning (GSL) methods have been proposed to learn graph structures and downstream tasks simultaneously, existing methods are predominantly designed for homogeneous graphs, while GSL for heterogeneous graphs remains largely unexplored. Two challenges arise in this context. Firstly, the quality of the input graph structure has a more profound impact on GNN-based heterogeneous GRL models compared to their homogeneous counterparts. Secondly, most existing homogenous GRL models encounter memory consumption issues when applied directly to heterogeneous graphs. In this paper, we propose a novel Graph Topology learning Enhanced Heterogeneous Graph Representation Learning framework (ToGRL).ToGRL learns high-quality graph structures and representations for downstream tasks by incorporating task-relevant latent topology information. Specifically, a novel GSL module is first proposed to extract downstream task-related topology information from a raw graph structure and project it into topology embeddings. These embeddings are utilized to construct a new graph with smooth graph signals. This two-stage approach to GSL separates the optimization of the adjacency matrix from node representation learning to reduce memory consumption. Following this, a representation learning module takes the new graph as input to learn embeddings for downstream tasks. ToGRL also leverages prompt tuning to better utilize the knowledge embedded in learned representations, thus enhancing adaptability to downstream tasks. Extensive experiments on five real-world datasets show that our ToGRL outperforms state-of-the-art methods by a large margin.

---

### FoleyDesigner: Immersive Stereo Foley Generation with Precise Spatio-Temporal Alignment for Film Clips

Authors: Mengtian Li, Kunyan Dai, Yi Ding, Ruobing Ni, Ying Zhang, Wenwu Wang, Zhifeng Xie
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05731v1

Foley art plays a pivotal role in enhancing immersive auditory experiences in film, yet manual creation of spatio-temporally aligned audio remains labor-intensive. We propose FoleyDesigner, a novel framework inspired by professional Foley workflows, integrating film clip analysis, spatio-temporally controllable Foley generation, and professional audio mixing capabilities. FoleyDesigner employs a multi-agent architecture for precise spatio-temporal analysis. It achieves spatio-temporal alignment through latent diffusion models trained on spatio-temporal cues extracted from video frames, combined with large language model (LLM)-driven hybrid mechanisms that emulate post-production practices in film industry. To address the lack of high-quality stereo audio datasets in film, we introduce FilmStereo, the first professional stereo audio dataset containing spatial metadata, precise timestamps, and semantic annotations for eight common Foley categories. For applications, the framework supports interactive user control while maintaining seamless integration with professional pipelines, including 5.1-channel Dolby Atmos systems compliant with ITU-R BS.775 standards, thereby offering extensive creative flexibility. Extensive experiments demonstrate that our method achieves superior spatio-temporal alignment compared to existing baselines, with seamless compatibility with professional film production standards. The project page is available at https://gekiii996.github.io/FoleyDesigner/ .

---

### Controllable Image Generation with Composed Parallel Token Prediction

Authors: Jamie Stirling, Noura Al-Moubayed, Chris G. Willcocks, Hubert P. H. Shum
Published: 2026-04-07
Categories: cs.LG
arXiv: https://arxiv.org/abs/2604.05730v1

Conditional discrete generative models struggle to faithfully compose multiple input conditions. To address this, we derive a theoretically-grounded formulation for composing discrete probabilistic generative processes, with masked generation (absorbing diffusion) as a special case. Our formulation enables precise specification of novel combinations and numbers of input conditions that lie outside the training data, with concept weighting enabling emphasis or negation of individual conditions. In synergy with the richly compositional learned vocabulary of VQ-VAE and VQ-GAN, our method attains a $63.4\%$ relative reduction in error rate compared to the previous state-of-the-art, averaged across 3 datasets (positional CLEVR, relational CLEVR and FFHQ), simultaneously obtaining an average absolute FID improvement of $-9.58$. Meanwhile, our method offers a $2.3\times$ to $12\times$ real-time speed-up over comparable methods, and is readily applied to an open pre-trained discrete text-to-image model for fine-grained control of text-to-image generation.

---

### Single-Stage Signal Attenuation Diffusion Model for Low-Light Image Enhancement and Denoising

Authors: Ying Liu, Junchao Zhang, Caiyun Wu
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05727v1

Diffusion models excel at image restoration via probabilistic modeling of forward noise addition and reverse denoising, and their ability to handle complex noise while preserving fine details makes them well-suited for Low-Light Image Enhancement (LLIE). Mainstream diffusion based LLIE methods either adopt a two-stage pipeline or an auxiliary correction network to refine U-Net outputs, which severs the intrinsic link between enhancement and denoising and leads to suboptimal performance owing to inconsistent optimization objectives. To address these issues, we propose the Signal Attenuation Diffusion Model (SADM), a novel diffusion process that integrates the signal attenuation mechanism into the diffusion pipeline, enabling simultaneous brightness adjustment and noise suppression in a single stage. Specifically, the signal attenuation coefficient simulates the inherent signal attenuation of low-light degradation in the forward noise addition process, encoding the physical priors of low-light degradation to explicitly guide reverse denoising toward the concurrent optimization of brightness recovery and noise suppression, thereby eliminating the need for extra correction modules or staged training relied on by existing methods. We validate that our design maintains consistency with Denoising Diffusion Implicit Models(DDIM) via multi-scale pyramid sampling, balancing interpretability, restoration quality, and computational efficiency.

---

### Beyond Semantics: Disentangling Information Scope in Sparse Autoencoders for CLIP

Authors: Yusung Ro, Jaehyun Choi, Junmo Kim
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05724v1

Sparse Autoencoders (SAEs) have emerged as a powerful tool for interpreting the internal representations of CLIP vision encoders, yet existing analyses largely focus on the semantic meaning of individual features. We introduce information scope as a complementary dimension of interpretability that characterizes how broadly an SAE feature aggregates visual evidence, ranging from localized, patch-specific cues to global, image-level signals. We observe that some SAE features respond consistently across spatial perturbations, while others shift unpredictably with minor input changes, indicating a fundamental distinction in their underlying scope. To quantify this, we propose the Contextual Dependency Score (CDS), which separates positionally stable local scope features from positionally variant global scope features. Our experiments show that features of different information scopes exert systematically different influences on CLIP's predictions and confidence. These findings establish information scope as a critical new axis for understanding CLIP representations and provide a deeper diagnostic view of SAE-derived features.

---

### GaussianGrow: Geometry-aware Gaussian Growing from 3D Point Clouds with Text Guidance

Authors: Weiqi Zhang, Junsheng Zhou, Haotian Geng, Kanle Shi, Shenkun Xu, Yi Fang, Yu-Shen Liu
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05721v1

3D Gaussian Splatting has demonstrated superior performance in rendering efficiency and quality, yet the generation of 3D Gaussians still remains a challenge without proper geometric priors. Existing methods have explored predicting point maps as geometric references for inferring Gaussian primitives, while the unreliable estimated geometries may lead to poor generations. In this work, we introduce GaussianGrow, a novel approach that generates 3D Gaussians by learning to grow them from easily accessible 3D point clouds, naturally enforcing geometric accuracy in Gaussian generation. Specifically, we design a text-guided Gaussian growing scheme that leverages a multi-view diffusion model to synthesize consistent appearances from input point clouds for supervision. To mitigate artifacts caused by fusing neighboring views, we constrain novel views generated at non-preset camera poses identified in overlapping regions across different views. For completing the hard-to-observe regions, we propose to iteratively detect the camera pose by observing the largest un-grown regions in point clouds and inpainting them by inpainting the rendered view with a pretrained 2D diffusion model. The process continues until complete Gaussians are generated. We extensively evaluate GaussianGrow on text-guided Gaussian generation from synthetic and even real-scanned point clouds. Project Page: https://weiqi-zhang.github.io/GaussianGrow

---

### Hackers or Hallucinators? A Comprehensive Analysis of LLM-Based Automated Penetration Testing

Authors: Jiaren Peng, Zeqin Li, Chang You, Yan Wang, Hanlin Sun, Xuan Tian, Shuqiao Zhang, Junyi Liu, Jianguo Zhao, Renyang Liu, Haoran Ou, Yuqiang Sun et al.
Published: 2026-04-07
Categories: cs.CR, cs.AI, cs.SE
arXiv: https://arxiv.org/abs/2604.05719v1

The rapid advancement of Large Language Models (LLMs) has created new opportunities for Automated Penetration Testing (AutoPT), spawning numerous frameworks aimed at achieving end-to-end autonomous attacks. However, despite the proliferation of related studies, existing research generally lacks systematic architectural analysis and large-scale empirical comparisons under a unified benchmark. Therefore, this paper presents the first Systematization of Knowledge (SoK) focusing on the architectural design and comprehensive empirical evaluation of current LLM-based AutoPT frameworks. At systematization level, we comprehensively review existing framework designs across six dimensions: agent architecture, agent plan, agent memory, agent execution, external knowledge, and benchmarks. At empirical level, we conduct large-scale experiments on 13 representative open-source AutoPT frameworks and 2 baseline frameworks utilizing a unified benchmark. The experiments consumed over 10 billion tokens in total and generated more than 1,500 execution logs, which were manually reviewed and analyzed over four months by a panel of more than 15 researchers with expertise in cybersecurity. By investigating the latest progress in this rapidly developing field, we provide researchers with a structured taxonomy to understand existing LLM-based AutoPT frameworks and a large-scale empirical benchmark, along with promising directions for future research.

---

### MPM: Mutual Pair Merging for Efficient Vision Transformers

Authors: Simon Ravé, Pejman Rasti, David Rousseau
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05718v1

Decreasing sequence length is a common way to accelerate transformers, but prior token reduction work often targets classification and reports proxy metrics rather than end-to-end latency. For semantic segmentation, token reduction is further constrained by the need to reconstruct dense, pixel-aligned features, and on modern accelerators the overhead of computing merge maps can erase expected gains. We propose Mutual Pair Merging (MPM), a training-free token aggregation module that forms mutual nearest-neighbor pairs in cosine space, averages each pair, and records a merge map enabling a gather-based reconstruction before the decoder so that existing segmentation heads can be used unchanged. MPM introduces no learned parameters and no continuous compression knob (no keep-rate or threshold). The speed-accuracy trade-off is set by a discrete insertion schedule. We benchmark end-to-end latency on an NVIDIA H100 GPU (with and without FlashAttention-2) and a Raspberry Pi 5 across standard segmentation datasets. On ADE20K, MPM reduces per-image latency by up to 60% for ViT-Tiny on Raspberry Pi 5, and increases throughput by up to 20% on H100 with FlashAttention-2 while keeping the mIoU drop below 3%. These results suggest that simple, reconstruction-aware, training-free token merging can translate into practical wall-clock gains for segmentation when overhead is explicitly accounted for.

---

### Can Large Language Models Reinvent Foundational Algorithms?

Authors: Jian Zhao, Haoren Luo, Yu Wang, Yuhan Cao, Pingyue Sheng, Tianxing He
Published: 2026-04-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.05716v1

LLMs have shown strong potential to advance scientific discovery. Whether they possess the capacity for foundational innovation, however, remains an open question. In this work, we focus on a prerequisite for foundational innovation: can LLMs reinvent foundational algorithms in computer science? Our \textit{Unlearn-and-Reinvent} pipeline applies LLM unlearning to remove a specific foundational algorithm, such as Dijkstra's or Euclid's algorithm, from an LLM's pretrained knowledge, and then tests whether the model can reinvent it in a controlled environment. To enable effective unlearning, we adopt a GRPO-based, on-policy unlearning method. Across 10 target algorithms, 3 strong open-weight models, and 3 hint levels, our experiments demonstrate that (1) the strongest model Qwen3-4B-Thinking-2507 successfully reinvents 50% of the algorithms with no hint, 70% at hint level 1, and 90% at hint level 2; (2) a few high-level hints can enhance the reinvention success rate, but even step-by-step hints fail for those complicated algorithms; and (3) test-time reinforcement learning enables successful reinvention for the Strassen algorithm at hint level 2. Through analyses of output trajectories and ablation studies, we find that generative verifier in the reinvention phase plays a critical role in sustaining models' reasoning strength, helping to avoid the ``thought collapse'' phenomenon. These findings offer insights into both the potential and current limits of LLMs' innovative thinking.

---

### In Depth We Trust: Reliable Monocular Depth Supervision for Gaussian Splatting

Authors: Wenhui Xiao, Ethan Goan, Rodrigo Santa Cruz, David Ahmedt-Aristizabal, Olivier Salvado, Clinton Fookes, Leo Lebrat
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05715v1

Using accurate depth priors in 3D Gaussian Splatting helps mitigate artifacts caused by sparse training data and textureless surfaces. However, acquiring accurate depth maps requires specialized acquisition systems. Foundation monocular depth estimation models offer a cost-effective alternative, but they suffer from scale ambiguity, multi-view inconsistency, and local geometric inaccuracies, which can degrade rendering performance when applied naively. This paper addresses the challenge of reliably leveraging monocular depth priors for Gaussian Splatting (GS) rendering enhancement. To this end, we introduce a training framework integrating scale-ambiguous and noisy depth priors into geometric supervision. We highlight the importance of learning from weakly aligned depth variations. We introduce a method to isolate ill-posed geometry for selective monocular depth regularization, restricting the propagation of depth inaccuracies into well-reconstructed 3D structures. Extensive experiments across diverse datasets show consistent improvements in geometric accuracy, leading to more faithful depth estimation and higher rendering quality across different GS variants and monocular depth backbones tested.

---

### SemLink: A Semantic-Aware Automated Test Oracle for Hyperlink Verification using Siamese Sentence-BERT

Authors: Guan-Yan Yang, Wei-Ling Wen, Shu-Yuan Ku, Farn Wang, Kuo-Hui Yeh
Published: 2026-04-07
Categories: cs.SE, cs.AI, cs.CL, cs.IR
arXiv: https://arxiv.org/abs/2604.05711v1

Web applications rely heavily on hyperlinks to connect disparate information resources. However, the dynamic nature of the web leads to link rot, where targets become unavailable, and more insidiously, semantic drift, where a valid HTTP 200 connection exists, but the target content no longer aligns with the source context. Traditional verification tools, which primarily function as crash oracles by checking HTTP status codes, often fail to detect semantic inconsistencies, thereby compromising web integrity and user experience. While Large Language Models (LLMs) offer semantic understanding, they suffer from high latency, privacy concerns, and prohibitive costs for large-scale regression testing. In this paper, we propose SemLink, a novel automated test oracle for semantic hyperlink verification. SemLink leverages a Siamese Neural Network architecture powered by a pre-trained Sentence-BERT (SBERT) backbone to compute the semantic coherence between a hyperlink's source context (anchor text, surrounding DOM elements, and visual features) and its target page content. To train and evaluate our model, we introduce the Hyperlink-Webpage Positive Pairs (HWPPs) dataset, a rigorously constructed corpus of over 60,000 semantic pairs. Our evaluation demonstrates that SemLink achieves a Recall of 96.00%, comparable to state-of-the-art LLMs (GPT-5.2), while operating approximately 47.5 times faster and requiring significantly fewer computational resources. This work bridges the gap between traditional syntactic checkers and expensive generative AI, offering a robust and efficient solution for automated web quality assurance.

---

### Untargeted analysis of volatile markers of post-exercise fat oxidation in exhaled breath

Authors: André Homeyer, Júlia Blanka Sziládi, Jan-Philipp Redlich, Jonathan Beauchamp, Y Lan Pham
Published: 2026-04-07
Categories: physics.med-ph, cs.LG
arXiv: https://arxiv.org/abs/2604.05707v1

Breath acetone represents a promising non-invasive biomarker for monitoring fat oxidation during exercise. However, its utility is limited by confounding factors, as well as by the fact that significant changes in concentration occur only hours post-exercise, which makes real-time assessment difficult. We performed an untargeted screening for volatile organic compounds (VOCs) that could serve as markers of fat oxidation beyond acetone, and investigated whether breath measurements taken during exercise could predict post-exercise changes in fat oxidation. Nineteen participants completed two 25-min cycling sessions separated by a brief 5-min rest period. VOC emissions were analysed using proton-transfer-reaction time-of-flight mass spectrometry (PTR-TOF-MS) during exercise and after a 90-min recovery period. Blood $β$-hydroxybutyrate (BOHB) concentrations served as the reference marker for fat oxidation. Among 773 relevant analytical features detected in the PTR-TOF-MS measurements, only four signals exhibited strong correlations with BOHB ($ρ$ $\geq$ 0.82, p = 0.0002)-all attributable to acetone or its isotopologues or fragments. End-of-exercise measurements of these signals enabled accurate prediction of participants with substantial post-exercise BOHB changes (F1 score $\geq$ 0.83, accuracy = 0.89). Our study did not reveal any novel breath-based biomarkers of fat oxidation, but it confirmed acetone as the key marker. Moreover, our findings suggest that breath acetone measurements during exercise may already enable basic predictions of post-exercise fat oxidation.

---

### QA-MoE: Towards a Continuous Reliability Spectrum with Quality-Aware Mixture of Experts for Robust Multimodal Sentiment Analysis

Authors: Yitong Zhu, Yuxuan Jiang, Guanxuan Jiang, Bojing Hou, Peng Yuan Zhou, Ge Lin Kan, Yuyang Wang
Published: 2026-04-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.05704v2

Multimodal Sentiment Analysis (MSA) aims to infer human sentiment from textual, acoustic, and visual signals. In real-world scenarios, however, multimodal inputs are often compromised by dynamic noise or modality missingness. Existing methods typically treat these imperfections as discrete cases or assume fixed corruption ratios, which limits their adaptability to continuously varying reliability conditions. To address this, we first introduce a Continuous Reliability Spectrum to unify missingness and quality degradation into a single framework. Building on this, we propose QA-MoE, a Quality-Aware Mixture-of-Experts framework that quantifies modality reliability via self-supervised aleatoric uncertainty. This mechanism explicitly guides expert routing, enabling the model to suppress error propagation from unreliable signals while preserving task-relevant information. Extensive experiments indicate that QA-MoE achieves competitive or state-of-the-art performance across diverse degradation scenarios and exhibits a promising One-Checkpoint-for-All property in practice.

---

### Dialogue Act Patterns in GenAI-Mediated L2 Oral Practice: A Sequential Analysis of Learner-Chatbot Interactions

Authors: Liqun He, Shijun, Chen, Mutlu Cukurova, Manolis Mavrikis
Published: 2026-04-07
Categories: cs.CL, cs.HC
arXiv: https://arxiv.org/abs/2604.05702v1

While generative AI (GenAI) voice chatbots offer scalable opportunities for second language (L2) oral practice, the interactional processes related to learners' gains remain underexplored. This study investigates dialogue act (DA) patterns in interactions between Grade 9 Chinese English as a foreign language (EFL) learners and a GenAI voice chatbot over a 10-week intervention. Seventy sessions from 12 students were annotated by human coders using a pedagogy-informed coding scheme, yielding 6,957 coded DAs. DA distributions and sequential patterns were compared between high- and low-progress sessions. At the DA level, high-progress sessions showed more learner-initiated questions, whereas low-progress sessions exhibited higher rates of clarification-seeking, indicating greater comprehension difficulty. At the sequential level, high-progress sessions were characterised by more frequent prompting-based corrective feedback sequences, consistently positioned after learner responses, highlighting the role of feedback type and timing in effective interaction. Overall, these findings underscore the value of a dialogic lens in GenAI chatbot design, contribute a pedagogy-informed DA coding framework, and inform the design of adaptive GenAI chatbots for L2 education.

---

### Optimal-Transport-Guided Functional Flow Matching for Turbulent Field Generation in Hilbert Space

Authors: Li Kunpeng, Wan Chenguang, Qu Zhisong, Lim Kyungtak, Virginie Grandgirard, Xavier Garbet, Yu Hua, Ong Yew Soon
Published: 2026-04-07
Categories: cs.LG
arXiv: https://arxiv.org/abs/2604.05700v1

High-fidelity modeling of turbulent flows requires capturing complex spatiotemporal dynamics and multi-scale intermittency, posing a fundamental challenge for traditional knowledge-based systems. While deep generative models, such as diffusion models and Flow Matching, have shown promising performance, they are fundamentally constrained by their discrete, pixel-based nature. This limitation restricts their applicability in turbulence computing, where data inherently exists in a functional form. To address this gap, we propose Functional Optimal Transport Conditional Flow Matching (FOT-CFM), a generative framework defined directly in infinite-dimensional function space. Unlike conventional approaches defined on fixed grids, FOT-CFM treats physical fields as elements of an infinite-dimensional Hilbert space, and learns resolution-invariant generative dynamics directly at the level of probability measures. By integrating Optimal Transport (OT) theory, we construct deterministic, straight-line probability paths between noise and data measures in Hilbert space. This formulation enables simulation-free training and significantly accelerates the sampling process. We rigorously evaluate the proposed system on a diverse suite of chaotic dynamical systems, including the Navier-Stokes equations, Kolmogorov Flow, and Hasegawa-Wakatani equations, all of which exhibit rich multi-scale turbulent structures. Experimental results demonstrate that FOT-CFM achieves superior fidelity in reproducing high-order turbulent statistics and energy spectra compared to state-of-the-art baselines.

---

### Let Geometry GUIDE: Layer-wise Unrolling of Geometric Priors in Multimodal LLMs

Authors: Chongyu Wang, Ting Huang, Chunyu Sun, Xinyu Ning, Di Wang, Hao Tang
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05695v2

Multimodal Large Language Models (MLLMs) have achieved remarkable progress in 2D visual tasks but still struggle to understand physical space in real-world visual streams. Recently, feed-forward geometric foundation models that implicitly extract geometric priors from visual inputs have offered a promising direction. However, many existing geometry-aware MLLMs extract features from a single deep encoder layer and perform one-shot fusion at the language-model input, potentially overlooking complementary information across geometric encoder layers, limiting access to fine-grained geometric cues, and hindering progressive cross-modal adaptation. To address these limitations, we propose GUIDE (Geometric Unrolling Inside MLLM Early-layers), a progressive framework for integrating geometric priors. GUIDE samples and aligns multi-level features from the geometric encoder and sequentially injects them into the early layers of the MLLM. This design enables the MLLM to continuously access multi-granularity geometric cues and progressively integrate them with visual representations during layer-wise semantic abstraction. GUIDE further introduces a dual context-aware gating mechanism at both the token and layer levels, selectively regulating geometric information to reduce redundant injection and interference with pretrained representations. Extensive experiments on spatial reasoning and 3D scene understanding benchmarks, including VSI-Bench, ScanRefer, and Scan2Cap, validate the effectiveness and cross-task applicability of GUIDE. Our 5B and 9B models achieve average scores of 71.5 and 72.1 on VSI-Bench, respectively.

---

### Let Geometry GUIDE: Layer-wise Unrolling of Geometric Priors in Multimodal LLMs

Authors: Chongyu Wang, Ting Huang, Chunyu Sun, Xinyu Ning, Di Wang, Hao Tang
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05695v1

Multimodal Large Language Models (MLLMs) have achieved remarkable progress in 2D visual tasks but still exhibit limited physical spatial awareness when processing real-world visual streams. Recently, feed-forward geometric foundation models, which implicitly extract geometric priors, have provided a new pathway to address this issue. However, existing geometry-aware MLLMs are predominantly constrained by the paradigm of single deep-layer extraction and input-level fusion. This flattened fusion leads to the loss of local geometric details and causes semantic mismatches in the early layers. To break this bottleneck, we propose GUIDE (Geometric Unrolling Inside MLLM Early-layers), a progressive geometric priors injection framework. GUIDE performs multi-level sampling within the geometric encoder, comprehensively capturing multi-granularity features ranging from local edges to global topologies. Subsequently, we rigorously align and fuse these multi-level geometric priors step-by-step with the early layers of the MLLM. Building upon the injection of multi-granularity geometric information, this design guides the model to progressively learn the 2D-to-3D transitional process. Furthermore, we introduce a context-aware gating that enables the model to fetch requisite spatial cues based on current semantics, thereby maximizing the utilization efficiency of spatial priors and effectively suppressing redundant geometric noise. Extensive experiments demonstrate that GUIDE significantly outperforms existing baselines on multiple complex spatial reasoning and perception tasks, establishing a novel paradigm for integrating 3D geometric priors into large models.

---

### CRFT: Consistent-Recurrent Feature Flow Transformer for Cross-Modal Image Registration

Authors: Xuecong Liu, Mengzhu Ding, Zixuan Sun, Zhang Li, Xichao Teng
Published: 2026-04-07
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2604.05689v1

We present Consistent-Recurrent Feature Flow Transformer (CRFT), a unified coarse-to-fine framework based on feature flow learning for robust cross-modal image registration. CRFT learns a modality-independent feature flow representation within a transformer-based architecture that jointly performs feature alignment and flow estimation. The coarse stage establishes global correspondences through multi-scale feature correlation, while the fine stage refines local details via hierarchical feature fusion and adaptive spatial reasoning. To enhance geometric adaptability, an iterative discrepancy-guided attention mechanism with a Spatial Geometric Transform (SGT) recurrently refines the flow field, progressively capturing subtle spatial inconsistencies and enforcing feature-level consistency. This design enables accurate alignment under large affine and scale variations while maintaining structural coherence across modalities. Extensive experiments on diverse cross-modal datasets demonstrate that CRFT consistently outperforms state-of-the-art registration methods in both accuracy and robustness. Beyond registration, CRFT provides a generalizable paradigm for multimodal spatial correspondence, offering broad applicability to remote sensing, autonomous navigation, and medical imaging. Code and datasets are publicly available at https://github.com/NEU-Liuxuecong/CRFT.

---

### Attention Editing: A Versatile Framework for Cross-Architecture Attention Conversion

Authors: Zhen Cheng, Hao-Bo Yang, Wan-Yi Huang, Jin-Long Li
Published: 2026-04-07
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2604.05688v1

Key-Value (KV) cache memory and bandwidth increasingly dominate large language model inference cost in long-context and long-generation regimes. Architectures such as multi-head latent attention (MLA) and hybrid sliding-window attention (SWA) can alleviate this bound, but integrating them into existing models remains difficult. Prior methods impose fine-grained structural requirements on both source and target attention modules, which cannot meet the feasible requirement in practical deployment. We present Attention Editing, a practical framework for converting already-trained large language models (LLMs) with new attention architectures without re-pretraining from scratch. Attention editing replaces the original attention with a learnable target module and trains it using progressive distillation, consisting of (1) layer-wise teacher-forced optimization with intermediate activation supervision to prevent cold-start error accumulation, and (2) model-level distillation on next-token distributions, optionally regularized by weak feature matching. We instantiate the framework on two different target--MLA and GateSWA, a gated hybrid SWA design, and apply it to Qwen3-8B and Qwen3-30B-A3B. The resulting models maintain competitive performance while delivering substantial efficiency improvements, demonstrating that large-scale attention conversion is both feasible and robust. Notably, experiments are conducted on an Ascend 910B clusters, offering a practical training case study on domestic hardware.

---

### 3D Smoke Scene Reconstruction Guided by Vision Priors from Multimodal Large Language Models

Authors: Xinye Zheng, Fei Wang, Yiqi Nie, Kun Li, Junjie Chen, Jiaqi Zhao, Yanyan Wei, Zhiliang Wu
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05687v2

Reconstructing 3D scenes from smoke-degraded multi-view images is particularly difficult because smoke introduces strong scattering effects, view-dependent appearance changes, and severe degradation of cross-view consistency. To address these issues, we propose a framework that integrates visual priors with efficient 3D scene modeling. We employ Nano-Banana-Pro to enhance smoke-degraded images and provide clearer visual observations for reconstruction and develop Smoke-GS, a medium-aware 3D Gaussian Splatting framework for smoke scene reconstruction and restoration-oriented novel view synthesis. Smoke-GS models the scene using explicit 3D Gaussians and introduces a lightweight view-dependent medium branch to capture direction-dependent appearance variations caused by smoke. Our method preserves the rendering efficiency of 3D Gaussian Splatting while improving robustness to smoke-induced degradation. Results demonstrate the effectiveness of our method for generating consistent and visually clear novel views in challenging smoke environments.

---

### LUDOBENCH: Evaluating LLM Behavioural Decision-Making Through Spot-Based Board Game Scenarios in Ludo

Authors: Ojas Jain, Dhruv Kumar
Published: 2026-04-07
Categories: cs.AI, cs.CL, cs.GT, cs.LG, cs.MA
arXiv: https://arxiv.org/abs/2604.05681v1

We introduce LudoBench, a benchmark for evaluating LLM strategic reasoning in Ludo, a stochastic multi-agent board game whose dice mechanics, piece capture, safe-square navigation, and home-path progression introduce meaningful planning complexity. LudoBench comprises 480 handcrafted spot scenarios across 12 behaviorally distinct decision categories, each isolating a specific strategic choice. We additionally contribute a fully functional 4-player Ludo simulator supporting Random, Heuristic, Game-Theory, and LLM agents. The game-theory agent uses Expectiminimax search with depth-limited lookahead to provide a principled strategic ceiling beyond greedy heuristics. Evaluating six models spanning four model families, we find that all models agree with the game-theory baseline only 40-46% of the time. Models split into distinct behavioral archetypes: finishers that complete pieces but neglect development, and builders that develop but never finish. Each archetype captures only half of the game theory strategy. Models also display measurable behavioral shifts under history-conditioned grudge framing on identical board states, revealing prompt-sensitivity as a key vulnerability. LudoBench provides a lightweight and interpretable framework for benchmarking LLM strategic reasoning under uncertainty. All code, the spot dataset (480 entries) and model outputs are available at https://anonymous.4open.science/r/LudoBench-5CBF/

---

### Intrinsic perturbation scale for certified oracle objectives with epigraphic information

Authors: Karim Bounja, Boujemaâ Achchab, Abdeljalil Sakat
Published: 2026-04-07
Categories: math.OC, cs.LG, math.FA
arXiv: https://arxiv.org/abs/2604.05678v1

We introduce a natural displacement control for minimizer sets of oracle objectives equipped with certified epigraphic information. Formally, we replace the usual local uniform value control of objective perturbations - uncertifiable from finite pointwise information without additional structure - by the strictly weaker requirement of a cylinder-localized vertical epigraphic control, naturally provided by certified envelopes. Under set-based quadratic growth (allowing nonunique minimizers), this yields the classical square-root displacement estimate with optimal exponent 1/2, without any extrinsic assumption.

---

### From Incomplete Architecture to Quantified Risk: Multimodal LLM-Driven Security Assessment for Cyber-Physical Systems

Authors: Shaofei Huang, Christopher M. Poskitt, Lwin Khin Shar
Published: 2026-04-07
Categories: cs.CR, cs.AI
arXiv: https://arxiv.org/abs/2604.05674v1

Cyber-physical systems often contend with incomplete architectural documentation or outdated information resulting from legacy technologies, knowledge management gaps, and the complexity of integrating diverse subsystems over extended operational lifecycles. This architectural incompleteness impedes reliable security assessment, as inaccurate or missing architectural knowledge limits the identification of system dependencies, attack surfaces, and risk propagation pathways. To address this foundational challenge, this paper introduces ASTRAL (Architecture-Centric Security Threat Risk Assessment using LLMs), an architecture-centric security assessment technique implemented in a prototype tool powered by multimodal LLMs. The proposed approach assists practitioners in reconstructing and analysing CPS architectures when documentation is fragmented or absent. By leveraging prompt chaining, few-shot learning, and architectural reasoning, ASTRAL extracts and synthesises system representations from disparate data sources. By integrating LLM reasoning with architectural modelling, our approach supports adaptive threat identification and quantitative risk estimation for cyber-physical systems. We evaluated the approach through an ablation study across multiple CPS case studies and an expert evaluation involving 14 experienced cybersecurity practitioners. Practitioner feedback suggests that ASTRAL is useful and reliable for supporting architecture-centric security assessment. Overall, the results indicate that the approach can support more informed cyber risk management decisions.

---

### Rectified Schrödinger Bridge Matching for Few-Step Visual Navigation

Authors: Wuyang Luan, Junhui Li, Weiguang Zhao, Wenjian Zhang, Tieru Wu, Rui Ma
Published: 2026-04-07
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2604.05673v3

Visual navigation is a core challenge in Embodied AI, requiring autonomous agents to translate high-dimensional sensory observations into continuous, long-horizon action trajectories. While generative policies based on diffusion models and Schrödinger Bridges (SB) effectively capture multimodal action distributions, they require dozens of integration steps due to high-variance stochastic transport, posing a critical barrier for real-time robotic control. We propose Rectified Schrödinger Bridge Matching (RSBM), a framework that exploits a shared velocity-field structure between standard Schrödinger Bridges ($\varepsilon=1$, maximum-entropy transport) and deterministic Optimal Transport ($\varepsilon\to 0$, as in Conditional Flow Matching), controlled by a single entropic regularization parameter $\varepsilon$. We prove two key results: (1) the conditional velocity field's functional form is invariant across the entire $\varepsilon$-spectrum (Velocity Structure Invariance), enabling a single network to serve all regularization strengths; and (2) reducing $\varepsilon$ linearly decreases the conditional velocity variance, enabling more stable coarse-step ODE integration. Anchored to a learned conditional prior that shortens transport distance, RSBM operates at an intermediate $\varepsilon$ that balances multimodal coverage and path straightness. Empirically, while standard bridges require $\geq 10$ steps to converge, RSBM achieves over 94% cosine similarity and 92% success rate in merely 3 integration steps -- without distillation or multi-stage training -- substantially narrowing the gap between high-fidelity generative policies and the low-latency demands of Embodied AI.

---

### Efficient machine unlearning with minimax optimality

Authors: Jingyi Xie, Linjun Zhang, Sai Li
Published: 2026-04-07
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2604.05669v1

There is a growing demand for efficient data removal to comply with regulations like the GDPR and to mitigate the influence of biased or corrupted data. This has motivated the field of machine unlearning, which aims to eliminate the influence of specific data subsets without the cost of full retraining. In this work, we propose a statistical framework for machine unlearning with generic loss functions and establish theoretical guarantees. For squared loss, especially, we develop Unlearning Least Squares (ULS) and establish its minimax optimality for estimating the model parameter of remaining data when only the pre-trained estimator, forget samples, and a small subsample of the remaining data are available. Our results reveal that the estimation error decomposes into an oracle term and an unlearning cost determined by the forget proportion and the forget model bias. We further establish asymptotically valid inference procedures without requiring full retraining. Numerical experiments and real-data applications demonstrate that the proposed method achieves performance close to retraining while requiring substantially less data access.

---

### CuraLight: Debate-Guided Data Curation for LLM-Centered Traffic Signal Control

Authors: Qing Guo, Xinhang Li, Junyu Chen, Zheng Guo, Shengzhe Xu, Lin Zhang, Lei Li
Published: 2026-04-07
Categories: cs.AI
arXiv: https://arxiv.org/abs/2604.05663v2

Traffic signal control (TSC) is a core component of intelligent transportation systems (ITS), aiming to reduce congestion, emissions, and travel time. Recent approaches based on reinforcement learning (RL) and large language models (LLMs) have improved adaptivity, but still suffer from limited interpretability, insufficient interaction data, and weak generalization to heterogeneous intersections. This paper proposes CuraLight, an LLM-centered framework where an RL agent assists the fine-tuning of an LLM-based traffic signal controller. The RL agent explores traffic environments and generates high-quality interaction trajectories, which are converted into prompt-response pairs for imitation fine-tuning. A multi-LLM ensemble deliberation system further evaluates candidate signal timing actions through structured debate, providing preference-aware supervision signals for training. Experiments conducted in SUMO across heterogeneous real-world networks from Jinan, Hangzhou, and Yizhuang demonstrate that CuraLight consistently outperforms state-of-the-art baselines, reducing average travel time by 5.34 percent, average queue length by 5.14 percent, and average waiting time by 7.02 percent. The results highlight the effectiveness of combining RL-assisted exploration with deliberation-based data curation for scalable and interpretable traffic signal control.

---

### SnapFlow: One-Step Action Generation for Flow-Matching VLAs via Progressive Self-Distillation

Authors: Wuyang Luan, Junhui Li, Weiguang Zhao, Wenjian Zhang, Tieru Wu, Rui Ma
Published: 2026-04-07
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2604.05656v1

Vision-Language-Action (VLA) models based on flow matching -- such as pi0, pi0.5, and SmolVLA -- achieve state-of-the-art generalist robotic manipulation, yet their iterative denoising, typically 10 ODE steps, introduces substantial latency: on a modern GPU, denoising alone accounts for 80% of end-to-end inference time. Naively reducing the step count is unreliable, degrading success on most tasks due to the velocity field being uncalibrated for single-step jumps. We present SnapFlow, a plug-and-play self-distillation method that compresses multi-step denoising into a single forward pass (1-NFE) for flow-matching VLAs. SnapFlow mixes standard flow-matching samples with consistency samples whose targets are two-step Euler shortcut velocities computed from the model's own marginal velocity predictions, avoiding the trajectory drift caused by conditional velocities, as we analyze theoretically. A zero-initialized target-time embedding lets the network switch between local velocity estimation and global one-step generation within a single architecture. SnapFlow requires no external teacher, no architecture changes, and trains in ~12h on a single GPU. We validate on two VLA architectures spanning a 6x parameter range, with identical hyperparameters: on pi0.5 (3B) across four LIBERO suites (40 tasks, 400 episodes), SnapFlow achieves 98.75% average success -- matching the 10-step teacher at 97.75% and slightly exceeding it -- with 9.6x denoising speedup and end-to-end latency reduced from 274ms to 83ms; on SmolVLA (500M), it reduces MSE by 8.3% with 3.56x end-to-end acceleration. An action-step sweep on long-horizon tasks reveals that SnapFlow maintains its advantage across execution horizons, achieving 93% at n_act=5 where the baseline reaches only 90%. SnapFlow is orthogonal to layer-distillation and token-pruning approaches, enabling compositional speedups.

---

### LLM Reasoning as Trajectories: Step-Specific Representation Geometry and Correctness Signals

Authors: Lihao Sun, Hang Dong, Bo Qiao, Qingwei Lin, Dongmei Zhang, Saravan Rajmohan
Published: 2026-04-07
Categories: cs.CL, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2604.05655v1

This work characterizes large language models' chain-of-thought generation as a structured trajectory through representation space. We show that mathematical reasoning traverses functionally ordered, step-specific subspaces that become increasingly separable with layer depth. This structure already exists in base models, while reasoning training primarily accelerates convergence toward termination-related subspaces rather than introducing new representational organization. While early reasoning steps follow similar trajectories, correct and incorrect solutions diverge systematically at late stages. This late-stage divergence enables mid-reasoning prediction of final-answer correctness with ROC-AUC up to 0.87. Furthermore, we introduce trajectory-based steering, an inference-time intervention framework that enables reasoning correction and length control based on derived ideal trajectories. Together, these results establish reasoning trajectories as a geometric lens for interpreting, predicting, and controlling LLM reasoning behavior.

---

### Multiscale Physics-Informed Neural Network for Complex Fluid Flows with Long-Range Dependencies

Authors: Prashant Kumar, Rajesh Ranjan
Published: 2026-04-07
Categories: physics.flu-dyn, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2604.05652v1

Fluid flows are governed by the nonlinear Navier-Stokes equations, which can manifest multiscale dynamics even from predictable initial conditions. Predicting such phenomena remains a formidable challenge in scientific machine learning, particularly regarding convergence speed, data requirements, and solution accuracy. In complex fluid flows, these challenges are exacerbated by long-range spatial dependencies arising from distant boundary conditions, which typically necessitate extensive supervision data to achieve acceptable results. We propose the Domain-Decomposed and Shifted Physics-Informed Neural Network (DDS-PINN), a framework designed to resolve such multiscale interactions with minimal supervision. By utilizing localized networks with a unified global loss, DDS-PINN captures global dependencies while maintaining local precision. The robustness of the approach is demonstrated across a suite of benchmarks, including a multiscale linear differential equation, the nonlinear Burgers' equation, and data-free Navier-Stokes simulations of flat-plate boundary layers. Finally, DDS-PINN is applied to the computationally challenging backward-facing step (BFS) problem; for laminar regimes (Re = 100), the model yields results comparable to computational fluid dynamics (CFD) without the need for any data, accurately predicting boundary layer thickness, separation, and reattachment lengths. For turbulent BFS flow at Re = 10,000, the framework achieves convergence to O(10^-4) using only 500 random supervision points (< 0.3 % of the total domain), outperforming established methods like Residual-based Attention-PINN in accuracy. This approach demonstrates strong potential for the super-resolution of complex turbulent flows from sparse experimental measurements.

---

### Probing Intrinsic Medical Task Relationships: A Contrastive Learning Perspective

Authors: Jonas Muth, Zdravko Marinov, Simon Reiß
Published: 2026-04-07
Categories: cs.CV
arXiv: https://arxiv.org/abs/2604.05651v1

While much of the medical computer vision community has focused on advancing performance for specific tasks, the underlying relationships between tasks, i.e., how they relate, overlap, or differ on a representational level, remain largely unexplored. Our work explores these intrinsic relationships between medical vision tasks, specifically, we investigate 30 tasks, such as semantic tasks (e.g., segmentation and detection), image generative tasks (e.g., denoising, inpainting, or colorization), and image transformation tasks (e.g., geometric transformations). Our goal is to probe whether a data-driven representation space can capture an underlying structure of tasks across a variety of 39 datasets from wildly different medical imaging modalities, including computed tomography, magnetic resonance, electron microscopy, X-ray ultrasound and more. By revealing how tasks relate to one another, we aim to provide insights into their fundamental properties and interconnectedness. To this end, we introduce Task-Contrastive Learning (TaCo), a contrastive learning framework designed to embed tasks into a shared representation space. Through TaCo, we map these heterogeneous tasks from different modalities into a joint space and analyze their properties: identifying which tasks are distinctly represented, which blend together, and how iterative alterations to tasks are reflected in the embedding space. Our work provides a foundation for understanding the intrinsic structure of medical vision tasks, offering a deeper understanding of task similarities and their interconnected properties in embedding spaces.

---

### See the Forest for the Trees: Loosely Speculative Decoding via Visual-Semantic Guidance for Efficient Inference of Video LLMs

Authors: Yicheng Ji, Jun Zhang, Jinpeng Chen, Cong Wang, Lidan Shou, Gang Chen, Huan Li
Published: 2026-04-07
Categories: cs.CL
arXiv: https://arxiv.org/abs/2604.05650v2

Video Large Language Models (Video-LLMs) excel in video understanding but suffer from high inference latency during autoregressive generation. Speculative Decoding (SD) mitigates this by applying a draft-and-verify paradigm, yet existing methods are constrained by rigid exact-match rules, severely limiting the acceleration potential. To bridge this gap, we propose LVSpec, the first training-free loosely SD framework tailored for Video-LLMs. Grounded in the insight that generation is governed by sparse visual-relevant anchors (mandating strictness) amidst abundant visual-irrelevant fillers (permitting loose verification), LVSpec employs a lightweight visual-relevant token identification scheme to accurately pinpoint the former. To further maximize acceptance, we augment this with a position-shift tolerant mechanism that effectively salvages positionally mismatched but semantically equivalent tokens. Experiments demonstrate that LVSpec achieves high fidelity and speed: it preserves >99.8 of target performance while accelerating Qwen2.5-VL-32B by 2.70x and LLaVA-OneVision-72B by 2.94x. Notably, it boosts the mean accepted length and speedup ratio by 136% and 35% compared to SOTA training-free SD methods for Video-LLMs.

---

### Analogical Reasoning as a Doctor: A Foundation Model for Gastrointestinal Endoscopy Diagnosis

Authors: Peixi Peng, Housheng Xie, Yanling Wei, Guangcong Ruan, Xiaoyang Zou, Qian Cao, Yongjian Nian, Guoyan Zheng
Published: 2026-04-07
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2604.05649v1

Gastrointestinal diseases impose a growing global health burden, and endoscopy is a primary tool for early diagnosis. However, routine endoscopic image interpretation still suffers from missed lesions and limited efficiency. Although AI-assisted diagnosis has shown promise, existing models often lack generalizability, adaptability, robustness, and scalability because of limited medical data, domain shift, and heterogeneous annotations. To address these challenges, we develop RATNet, a foundation model for gastrointestinal endoscopy imaging based on analogical reasoning. RATNet acquires and transfers knowledge from heterogeneous expert annotations across five gastrointestinal endoscopy datasets through a cyclic pre-training strategy. Its architecture consists of an encoder, a relevance-knowledge acquisition and transfer (RAT) module, a projector, and a multi-task head, and supports fine-tuning, linear probing, and zero-shot transfer. Evaluations show that RATNet outperforms existing foundation models, including GastroNet and GastroVision, across six scenarios: diagnosis of common gastrointestinal diseases, few-shot learning for rare diseases, zero-shot transfer to new medical sites, robustness under long-tailed disease distributions, adaptation to novel diseases, and privacy-preserving deployment via federated learning. Its advantage comes from an analogical reasoning mechanism that matches image-derived posterior knowledge to a learned prior knowledge base and transfers relative knowledge to guide diagnosis, improving generalization and resistance to bias. RATNet is open and cost-effective, supports automatic integration of heterogeneous annotations without manual label unification, and reduces data acquisition costs, making it a practical foundation for intelligent gastrointestinal diagnosis, especially in resource-limited settings.

---