Search arXiv⌕ Search

arXiv subjects

Yi Lin

Publications and source records attributed to Yi Lin.

At least 19 recordsLinked to original sources
arXiv2026

Efficient Multi-modal Video Fusion Network with Dynamic Region-Guided Temporal Interaction

Multi-modal video fusion aims to generate temporally coherent videos with salient targets and rich structural details by integrating complementary information from heterogeneous sensors. However, existing video fusion methods commonly rely on optical flow estimation and feature warping for inter-frame alignment and cross-frame information aggregation to maintain temporal consistency, thereby introducing considerable computational overhead and making the fusion performance vulnerable to flow estimation errors. To address these limitations, this paper proposes an efficient multi-modal video fusion network with dynamic region-guided temporal interaction. Specifically, the proposed spatio-temporal interaction module leverages the Test-Time Training paradigm to model temporal dependencies with linear complexity. Building on this design, the module restricts temporal modeling to key dynamic regions in the video sequence, as static background regions provide limited temporal cues. By excluding low-variation tokens from unnecessary temporal interaction, this selective strategy contributes to improved fusion quality and temporal consistency. In addition, leveraging the strong ability of pretrained MAE features to capture global structural dependencies, we introduce a feature-level temporal consistency constraint to suppress local flickering by aligning the inter-frame variation directions of the fused and source videos. We further propose a new MAE-based temporal metric that quantifies directional distance between their temporal variations within dynamic regions. Experimental results demonstrate that, compared with existing state-of-the-art methods, our method achieves superior fusion quality and temporal consistency on multiple benchmark datasets, while exhibiting high computational efficiency under edge-device deployment settings.

cs.CV↗
arXiv2026

Learning and Reasoning with Geometry: How Generation Helps Spatial Understanding

Image and video generators can produce spatially consistent visual content, but how to use this capability to improve spatial understanding remains underexplored. We study this problem from two perspectives: geometric supervision during training and explicit geometric evidence during inference. To this end, we introduce a unified geometric understanding and generation model (GeoUG) that integrates depth and multi-view generation with spatial understanding. First, we design a multi-scale depth codec that enables metric-depth generation using a pretrained visual autoencoder, and show that joint training yields task-dependent gains in spatial understanding. Second, we introduce 3DCueBench to control geometric difficulty while balancing selected 2D cues, and find that generated depth provides larger reasoning gains on questions requiring fine-grained geometric discrimination. Building on this finding, we develop adaptive interleaved reasoning, in which GeoUG first drafts an answer and then decides whether to generate geometric evidence to confirm or revise it. We train this decision through reinforcement learning, balancing accuracy gains over retaining the draft against generation cost. GeoUG achieves circular accuracies of 97.9%, 90.8%, and 94.2% on CV-Bench-3D, 3DCueBench, and DA-2K with fewer generation calls than always generating.

cs.CV↗
arXiv2026

TIM: Temporal Decoupling with Iterative Mutual-Refinement Model for Longitudinal Radiology Report Generation

Automatic radiology report generation (RRG) aims to translate medical images into diagnostic text, reducing radiologists' workload and standardizing clinical documentation. Nonetheless, existing approaches mainly focus on single-timepoint analysis and fail to capture temporal disease evolution across longitudinal examinations. While recent longitudinal RRG (LRRG) approaches incorporate historical data, they often combine images from different time points within a single representation space, leading to blurred semantics and inconsistent temporal reasoning. In this work, we propose a Temporal Decoupling with Iterative Mutual-Refinement Model (TIM), a two-stage framework that explicitly decouples spatial pathology from temporal progression and iteratively refines reports through mutual feedback. Stage I performs temporal-decoupled representation learning, separating temporal evolution patterns from disease-specific features and generating radiology reports for both prior and current studies. Stage II introduces a mutual report refinement mechanism that identifies diagnostic inconsistencies within prior reports and iteratively rectifies both prior and current reports through error-sensitive feedback. Experiments on the Longitudinal-MIMIC dataset demonstrate that TIM surpasses existing single-image and longitudinal baselines, achieving new state-of-the-art performance across both language and clinical metrics. Code is available at https://github.com/yihengd/TIM.

cs.CV↗
arXiv2026

Training Long-Context Vision-Language Models Effectively with Generalization Beyond 128K Context

Long-context modeling is becoming a core capability of modern large vision-language models (LVLMs), enabling sustained context management across long-document understanding, video analysis, and multi-turn tool use in long-horizon agents. Yet practical training recipes remain insufficiently explored, particularly for designing and balancing long-context data mixtures. In this work, we present a systematic study of long-context continued pre-training (LongPT) for LVLMs, extending a 7B model from 32K to 128K context with extensive ablations on long-document data. We first show that long-document VQA is substantially more effective than OCR transcription. Building on this observation, our ablations further yield three key findings: i) for sequence-length distribution, balanced data outperforms target-length-focused data (e.g., 128K), suggesting that long-context ability requires generalizable key-information retrieval across various lengths and positions; ii) retrieval remains the primary bottleneck, favoring retrieval-heavy mixtures with modest reasoning data for task diversity; iii) pure long-document VQA largely preserves short-context capabilities, suggesting that instruction-formatted long data reduces the need for short-data mixing. Based on these findings, we introduce MMProLong, obtained by long-context continued pre-training from Qwen2.5-VL-7B with only a 5B-token budget. MMProLong improves long-document VQA scores by 7.11% and maintains strong performance at 256K and 512K contexts beyond its 128K training window, without additional training. It further generalizes to webpage-based multimodal needle retrieval, long-context vision-text compression, and long-video understanding without task-specific supervision. Overall, our study establishes a practical LongPT recipe and an empirical foundation for advancing long-context vision-language models.

cs.CV↗
arXiv2026

CrossTimeEdit: A Decade-Spanning Cross-View Dataset and Reward-Guided Editing for Historical Street-View Generation

Historical street-view imagery records urban evolution, but uneven coverage leaves substantial gaps in historical records. Generating plausible past appearances requires restoring changed structures while preserving persistent scene content. We construct VIGOR-his, a decade-spanning cross-view dataset containing 43,653 location-level quadruplets across 11 cities on three continents. Its automated pipeline performs spatial pairing, consistency screening, change classification, and the generation and validation of satellite-based change descriptions and local editing instructions. Based on VIGOR-his, we propose CrossTimeEdit, a model that reformulates historical street-view generation as editing, using recent street views to constrain viewpoint and unchanged appearance and temporal satellite differences as change evidence. Starting from FLUX.2 [Klein] 4B, we train CrossTimeEdit through supervised fine-tuning (SFT) followed by online reinforcement learning (RL). We design three street-view editing criteria, namely Instruction Alignment (IA), Background Preservation (BP), and Quality and Physical Plausibility (QP), as both RL reward dimensions and evaluation metrics. We optimize this multi-reward objective using Within Group Relative Policy Optimization for flow-matching models (Flow-GRPO) with Group reward-Decoupled Normalization Policy Optimization (GDPO), which normalizes each reward dimension before aggregation. CrossTimeEdit improves overall performance across the three editing criteria by 17.12\% over the pretrained baseline and outperforms cross-view generation models in scene consistency, visual realism, and perceptual quality. The implementation code, dataset, and model weights are available at https://luhanwen67.github.io/CrossTimeEdit-release/.

cs.CV↗
arXiv2026

SaplingGuard: A Multidimensional-Profile-Aware Multi-Agent Guardrail for Developmentally Safe Adolescent-LLM Interaction

As adolescents increasingly use LLMs in everyday life, ensuring safe and developmentally appropriate responses has become essential. However, existing LLM guardrails primarily target explicit harmful content in isolated prompts or responses and are less effective at identifying implicit, context-dependent developmental risks. To address this limitation, we propose SaplingGuard, a plug-and-play, profile-aware and dialogue-aware guardrail that requires no modification to downstream model parameters. SaplingGuard decomposes adolescent safety intervention into three specialized agents for user profile construction, context-aware risk assessment, and intent-preserving prompt optimization. Together, these agents leverage the current prompt, preceding dialogue, and structured user characteristics to identify contextual risks and guide downstream response generation. We evaluate SaplingGuard on SaplingBench, which contains 276 three-turn dialogues spanning seven categories of developmental risk. Across ten adolescent profile conditions and nine open- and closed-source downstream LLMs, profile-aware retrieval improves the Major Hit rate from 50.8% to 63.7+/-1.1%. End-to-end intervention further reduces the average harmful response rate from 17.10% to 5.27% and increases the average safety score from 0.7017 to 1.0043. These results show that user-profile and dialogue context provide complementary signals for identifying implicit developmental risks, and that SaplingGuard can serve as an effective external safety layer for adolescent-LLM interaction.

cs.CR↗
arXiv2026

CXR-LT 2026 Challenge: Multi-Center Long-Tailed and Zero Shot Chest X-ray Classification

Chest X-ray (CXR) interpretation is hindered by the long-tailed distribution of pathologies and the open-world nature of clinical environments. Existing benchmarks often rely on closed-set classes from a single institution, failing to capture the prevalence of rare diseases or the appearance of novel findings. To address this, we present the CXR-LT challenge. The first event, CXR-LT 2023, established a large-scale benchmark for long-tailed multi-label CXR classification and identified key challenges in rare disease recognition. CXR-LT 2024 further expanded the label space and introduced a zero-shot task to study generalization to unseen findings. Building on the success of CXR-LT 2023 and 2024, this third iteration of the benchmark introduces a multi-center dataset comprising over 145,000 images from PadChest and NIH Chest X-ray datasets. Additionally, all development and test sets in CXR-LT 2026 are annotated by radiologists, providing a more reliable and clinically grounded evaluation than report-derived labels. The challenge defines two core tasks this year: (1) Robust Multi-Label Classification on 30 known classes and (2) Open-World Generalization to 6 unseen (out-of-distribution) rare disease classes. This paper summarizes the overview of the CXR-LT 2026 challenge. We describe the data collection and annotation procedures, analyze solution strategies adopted by participating teams, and evaluate head-versus-tail performance, calibration, and cross-center generalization gaps. Our results show that vision-language foundation models improve both in-distribution and zero-shot performance, but detecting rare findings under multi-center shift remains challenging. Our study provides a foundation for developing and evaluating AI systems in realistic long-tailed and open-world clinical conditions.

cs.CV↗
arXiv2026

DAGLFNet: Deep Feature Attention Guided Global and Local Feature Fusion for Pseudo-Image Point Cloud Segmentation

Environmental perception systems are crucial for high-precision mapping and autonomous navigation, with LiDAR serving as a core sensor providing accurate 3D point cloud data. Efficiently processing unstructured point clouds while extracting structured semantic information remains a significant challenge. In recent years, numerous pseudo-image-based representation methods have emerged to balance efficiency and performance by fusing 3D point clouds with 2D grids. However, the fundamental inconsistency between the pseudo-image representation and the original 3D information critically undermines 2D-3D feature fusion, posing a primary obstacle for coherent information fusion and leading to poor feature discriminability. This work proposes DAGLFNet, a pseudo-image-based semantic segmentation framework designed to extract discriminative features. It incorporates three key components: first, a Global-Local Feature Fusion Encoding (GL-FFE) module to enhance intra-set local feature correlation and capture global contextual information; second, a Multi-Branch Feature Extraction (MB-FE) network to capture richer neighborhood information and improve the discriminability of contour features; and third, a Feature Fusion via Deep Feature-guided Attention (FFDFA) mechanism to refine cross-channel feature fusion precision. Experimental evaluations demonstrate that DAGLFNet achieves mean Intersection-over-Union (mIoU) scores of 69.9% and 78.7% on the validation sets of SemanticKITTI and nuScenes, respectively. The method achieves an excellent balance between accuracy and efficiency.

cs.CV↗
arXiv2026

Equivariant Morse theory for Lie algebra actions on Riemannian foliations

We consider a transverse isometric action of a finite-dimensional Lie algebra $\mathfrak g$ on a Riemannian foliation. In this setting, we study equivariant Morse--Bott theory on the leaf space of the foliation. Among other results, we establish a foliated Morse--Bott lemma for $\mathfrak g$-invariant basic Morse--Bott functions and a foliated analogue of the usual handle presentation theorem. In the non-equivariant case, we use these results to give a new proof of the Morse inequalities for Riemannian foliations. In the equivariant case, we apply them to Hamiltonian actions of abelian Lie algebras on presymplectic manifolds whose underlying foliations are Riemannian, and we extend the Kirwan surjectivity and injectivity theorems from equivariant symplectic geometry to this setting. As a consequence, Kirwan surjectivity and injectivity hold for Hamiltonian torus actions on symplectic orbifolds.

math.DG↗
arXiv2026

Descent of Basic Forms to Quotients by Locally Free Lie Group Actions

We prove an equivariant version of the theorem of Hector, Mac\'ıas-Virgós, and Sanmart\'ın-Carbón identifying diffeological forms on the leaf space of a foliation with basic forms. For any group $K$ acting by foliation-preserving diffeomorphisms, we show that this identification is an isomorphism of $K$-equivariant cochain complexes. We then establish a descent theorem for basic differential forms under locally free Lie group actions without assuming properness. Let a Lie group $H$, not necessarily connected or second countable, act smoothly and locally freely on a second countable manifold $M$, and let $\mathcal F$ be the foliation by $H_0$-orbits. We prove that pullback induces an isomorphism \[ Ω^\bullet(M/H)\congΩ^\bullet(M,\mathcal F)^H \] provided that $H$ is second countable or that the induced action of $H/H_0$ on $M/H_0$ satisfies a natural subduction condition. We also give a smooth free action for which descent fails, showing that an additional hypothesis is genuinely necessary.

math.DG↗
arXiv2026

A Source-Grounded Framework for Constructing and Evaluating Progressive Multimodal Diagnostic Dialogues from Clinical Case Reports

Clinical diagnosis requires progressive integration of patient history, physical examination, laboratory findings, medical images, and diagnostic-informative tests. However, most multimodal medical benchmarks evaluate fixed inputs or endpoint answers, while fully interactive diagnostic agents conflate evidence selection with evidence interpretation. We present a source-grounded framework to construct progressive multimodal diagnostic dialogues from case reports and an evaluation strategy for assessing MLLMs on final diagnosis, diagnostic reasoning, and image-finding interpretation. Evaluation on 24 internal medicine case reports showed that our framework can accurately convert case reports into reference dialogues, achieving a diagnosis F1 of 0.99 and a reasoning-quality score of 4.79 out of 5. Evaluation on two frontier MLLMs (o4-mini and Claude Haiku 4.5) achieved reasoning-quality scores of 2.75 and 2.50, respectively, with substantially lower diagnosis, reasoning, and image-finding F1 scores. The results demonstrate that fluent responses do not necessarily reflect evidence-grounded clinical reasoning and highlight the utility of the proposed framework for evaluating multimodal diagnostic reasoning.

cs.CL↗
arXiv2026

LLM assisted writing deserves empirical evaluation

LLM-assisted writing is often treated as a detection problem, as it raises questions about clarity, integrity, equity, and evaluation. An analysis of 69,209 Health Informatics papers links it to more focused presentation, broader citation practices, and more globally distributed authorship. These patterns do not prove better science, but they support evaluating manuscripts by scholarly quality and accountability rather than by tool use.

cs.CL↗
arXiv2026

Evaluating Multi-Turn Multimodal Diagnostic Reasoning on Challenging Real-World Clinical Cases

Clinical diagnostic evaluation should not only assess whether models can provide correct diagnoses, but also reflect the realities of clinical practice, including progressive disclosure of multimodal information, dynamic updating of diagnostic hypotheses, and continuous refinement of clinical reasoning. However, existing evaluations of multimodal large language models (MLLMs) typically rely on single-turn or isolated tasks, making it difficult to fully capture the complexity of real-world clinical diagnosis. To bridge this gap, we developed ClinMM-Bench, the largest multi-turn multimodal clinical diagnostic evaluation benchmark to date. ClinMM-Bench contains 1,089 challenging real-world clinical cases and 3,760 medical images across eight specialties. We systematically evaluated 15 representative MLLMs using a two-level evaluation framework that assessed both diagnostic accuracy and diagnostic reasoning quality. Results showed that proprietary models achieved the highest overall diagnostic accuracy, but the proportion of completely correct diagnoses remained limited across all models. In terms of diagnostic reasoning quality, current models can identify plausible diagnostic directions but still have considerable limitations in generating reliable diagnostic reasoning. Error analysis further identified five representative failure modes: information synthesis failure, knowledge mapping error, perception error, premature closure, and visual hallucination.

cs.CL↗
arXiv2026

BrainFIBRE: A Foundation Model via Information Decomposition for Brain Microstructure

Diffusion MRI probes brain microstructure with particular sensitivity to early cerebrovascular and neurodegenerative changes. Neurite Orientation Dispersion and Density Imaging (NODDI) decomposes the diffusion signal into three biophysically interpretable maps: neurite density index (NDI), orientation dispersion index (ODI), and free water fraction (FWF), capturing neurite packing, fiber coherence, and extracellular fluid. These 3D maps offer a rich substrate for transferable microstructural representations, yet integrating them is challenging: standard representation learning struggles to disentangle the unique information in each map from their shared and synergistic interactions. We present BrainFIBRE, the first foundation model for brain microstructure, pretrained on NODDI-derived maps from 55,592 UK Biobank participants. We propose Self-supervised Partial Information Decomposition (SPID), which extends PID-guided multimodal learning to the self-supervised regime for the first time. A novel Counterfactual Candidate Construction (CCC) paradigm perturbs inter-modality alignment through modality dropping and swapping, providing the contrastive signal for a Mixture-of-Experts architecture to disentangle unique, synergistic, and redundant information without any downstream label. On both Caucasian and Asian cohorts, BrainFIBRE achieves state-of-the-art performance across diverse tasks predicting age, sex, cerebrovascular and neurodegenerative markers, and cognition, while yielding neurobiologically interpretable representations that reveal task- and cohort-specific interaction patterns. BrainFIBRE establishes a versatile foundation for neuroimaging analysis at the microstructural level.

cs.CV↗
arXiv2026

Prompt-Guided Foundation Model Tuning for Pathology Image Classification

Foundation models have become pivotal in advancing computational pathology, particularly for whole slide image (WSI) classification. However, prevailing methodologies often rely on frozen, pre-trained models for feature extraction, overlooking the pronounced domain shift and task discrepancy between the pre-training and downstream tasks. To address this challenge, we propose PAMT, a novel Prompt-guided Adaptive Model Transformation framework that enables precise adaptation of general foundation models to the distinct domain of histopathology. To encapsulate the intricate distributions characteristic of histopathological data, we introduce Representative Patch Sampling (RPS) and Prototypical Visual Prompt (PVP), which reconstruct the input into compact yet highly informative representations. Further, to effectively bridge the domain gap, we incorporate Adaptive Model Transformation (AMT) via adapter modules within the feature extraction pipeline, facilitating the acquisition of domain-specific features by the foundation model. We conduct rigorous evaluation across 14 publicly available datasets and demonstrate consistent, substantial improvements in classification accuracy. These results establish PAMT as a compelling new benchmark for pathology image classification and underscore the critical value of targeted model adaptation within computational pathology.

cs.CV↗
arXiv2026

A Transverse Averaging Operator and Cohomology of Quotients by Homogeneous Non-closed Subgroups

In this article, we introduce a transverse averaging operator for basic forms on a complete Riemannian foliation with compact leaf closure space, equipped with an isometric transverse Lie algebra action. In contrast to the classical averaging operator in equivariant geometry, which is defined by integration over a compact Lie group, our operator is constructed purely from infinitesimal transverse data and does not require any global group action. We prove that every closed basic form is sent to an invariant basic form representing the same basic cohomology class. The main application is formulated independently of foliation theory. We compute the diffeological de Rham cohomology of the homogeneous quotient $G/H$, where $G$ is a connected Lie group, not necessarily compact, and $H$ is a connected Lie subgroup, not necessarily closed. Let $\mathfrak g$ and $\mathfrak h$ be the Lie algebras of $G$ and $H$, respectively. Assuming that $\mathfrak g$ is of compact type and that $G/\overline{H}$ is compact, we prove that $ H^\bullet_{\mathrm{dR}}(G/H) \cong H^\bullet(\mathfrak g,\mathfrak h)$. When $\mathfrak h$ is an ideal in $\mathfrak g$, the compact-type assumption on $\mathfrak g$ can be dropped, and under the sole assumption that $G/\overline{H}$ is compact we obtain $ H^\bullet_{\mathrm{dR}}(G/H) \cong H^\bullet(\mathfrak g/\mathfrak h)$. These results extend the classical Chevalley--Eilenberg computation from compact Lie groups and closed subgroups to homogeneous quotients $G/H$ by possibly non-closed subgroups $H$.

math.DG↗
arXiv2026

MonteRET: AI Agent Enhancing Multimodal LLMs with Multi-granularity Knowledge Retrieval for Chest CT Report Generation

Automated chest CT report generation remains challenging because clinically faithful reporting requires both whole-volume understanding and accurate description of localized anatomical findings. Here we developed and retrospectively evaluated MonteRET, a region-aware retrieval-enhanced framework for generating chest CT findings sections. MonteRET integrates global CT features with region-level anatomical representations, retrieves clinically relevant knowledge using predicted medical conditions and region-level vision-language alignment, and refines initial reports through a knowledge-guided report rewriting agent. We trained our model on a public cohort with 24,128 CT scans from RadGenome-ChestCT. We evaluated MonteRET on the public RadGenome-ChestCT test set of 1,564 CT scans and an external cohort of 82 CT scans from NewYork-Presbyterian/Weill Cornell Medical Center. MonteRET improved report quality, semantic similarity, and clinical efficacy compared with a matched baseline and several state-of-the-art methods. Gains were most pronounced for recall, suggesting fewer omitted findings. Human expert evaluation by radiology residents also favored MonteRET.

cs.CV↗
arXiv2026

UP: Unbounded Positive Asymmetric Optimization for Breaking the Exploration-Stability Dilemma

Reinforcement learning (RL) has become the standard paradigm for enhancing the complex reasoning capabilities of large language models (LLMs). To achieve sample efficiency, modern RL frameworks rely on importance sampling (IS). However, these algorithms suffer from an exploration-stability dilemma. Pure IS often leads to catastrophic training instability, while standard clipping mechanisms used to mitigate this instability strictly constrain the policy update budget. By formalizing the concept of Probability Capacity (Cap), we reveal that conservative clipping structurally stifles exploration by prematurely truncating the update budget for correct but low-confidence reasoning paths. To break free from these constraints, we propose Unbounded Positive Asymmetric Optimization (UP), a universal and plug-and-play objective. UP theoretically restructures the optimization process by anchoring the policy to its current state via the stop-gradient operator. This asymmetric design unleashes unclipped, stable gradients for positive advantages to maximize exploration, while maintaining standard clipping safeguards for negative advantages to prevent training instability. Furthermore, our formulation readily extends across different optimization granularities, including token-level (GRPO, DAPO) and sequence-level (GSPO) frameworks. Extensive experiments demonstrate that UP enhances exploration capacity and achieves superior reasoning accuracy across diverse RL algorithms (DAPO, GSPO, and GRPO), model architectures (Dense, MoE, and vision-language), and training modalities (language and multimodal), validating UP as a truly universal plug-and-play enhancement for RL-based training.

cs.LG↗