arXiv subjects

Fei Su

Publications and source records attributed to Fei Su.

At least 19 recordsLinked to original sources

arXiv2026

H-OPD: Confidence Aware Heterogeneous Multi-Teacher Multimodal On-policy Distillation

On-policy distillation (OPD) has recently emerged as an effective post-training paradigm by providing supervision on student-generated trajectories. However, existing OPD methods for multimodal reasoning usually rely on a static teacher routing, assigning each sample to a single teacher based on modality or task type. This ignores that visual grounding and abstract reasoning may dominate different decoding steps, making a single teacher insufficient for the full trajectory. To this end, H-OPD is proposed as a confidence-aware heterogeneous multi-teacher OPD framework for multimodal reasoning. By verifying the complementarity of heterogeneous teachers in the same reasoning process, H-OPD replaces task or sample level teacher routing with token-level teacher arbitration along the shared student trajectory. H-OPD employs vision-to-language description transfer to enable text-only teachers to access key visual semantics, and uses a confidence-aware arbitration mechanism to dynamically combine vision-language teacher and text-only teachers at each token. Extensive evaluations over 11 widely-used reasoning benchmarks showcase the superior performance of our method.

cs.CV↗

arXiv2026

M2S-AVSR: Modality-aware Multi-view Self-supervised Representation for Robust Audio-Visual Speech Recognition

Audio-Visual Speech Recognition (AVSR) enhances speech recognition robustness by leveraging visual cues, while real-world scenarios remain challenging due to viewpoint variation, audio distortion, and visual occlusion, which degrade modality quality and increase audio-visual asynchrony. In this paper, we propose a novel Modality-aware Multi-view Self-supervised representation framework for robust Audio-Visual Speech Recognition (M2S-AVSR). First, we introduce a multi-view representation learning encoder to learn view-invariant visual speech representations. Next, we employ a modality-aware module that explicitly models modality quality and cross-modal synchrony to perform fine-grained modality-aware fusion, enabling fine-grained visual information injection during decoding. In addition, we release AISHELL8-RealScene, a public multi-scenario, multi-view conversational audio-visual dataset recorded in real-world environments, and establish a speech recognition benchmark on it. Experiments on English and Mandarin benchmarks demonstrate the effectiveness of the proposed method under challenging conditions. On LRS3, M2S-AVSR achieves up to 29.4% relative improvement under viewpoint perturbation and visual degradation settings. Our method also achieves new state-of-the-art performance on the MISP2021-AVSR test set. On AISHELL8-RealScene, it achieves the best result in outdoor scenes. The proposed method and dataset provide useful support for future research on robust speech and multimodal tasks under realistic conditions.

eess.AS↗

arXiv2026

CHRep: Cross-modal Histology Representation and Post-hoc Calibration for Spatial Gene Expression Prediction

Spatial transcriptomics (ST) enables spatially resolved gene profiling but remains expensive and low-throughput, limiting large-cohort studies and routine clinical use. Predicting spatial gene expression from routine hematoxylin and eosin (H&E) slides is a promising alternative, yet under realistic leave-one-slide-out evaluation, existing models often suffer from slide-level appearance shifts and regression-driven over-smoothing that suppress biologically meaningful variation. CHRep is a two-phase framework for robust histology-to-expression prediction. In the training phase, CHRep learns a structure-aware representation by jointly optimizing correlation-aware regression, symmetric image-expression alignment, and coordinate-induced spatial topology regularization. In the inference phase, cross-slide robustness is improved without backbone fine-tuning through a lightweight calibration module trained on the training slides, which combines a non-parametric estimate from a training gallery with a magnitude-regularized correction module. Unlike prior embedding-alignment or retrieval-based transfer methods that rely on a single prediction route, CHRep couples topology-preserving representation learning with post-hoc calibration, enabling stable neighborhood retrieval and controlled bias correction under slide-level shifts. Across the three cohorts, CHRep consistently improves gene-wise correlation under leave-one-slide-out evaluation, with the largest gains observed on Alex+10x. Relative to HAGE, the Pearson correlation coefficient on all considered genes [PCC(ACG)] increases by 4.0% on cSCC and 9.8% on HER2+. Relative to mclSTExp, PCC(ACG) further improves by 39.5% on Alex+10x, together with 9.7% and 9.0% reductions in mean squared error (MSE) and mean absolute error (MAE), respectively.

cs.CV↗

arXiv2026

PathGLS: Evaluating Pathology Vision-Language Models without Ground Truth through Multi-Dimensional Consistency

Vision-Language Models (VLMs) offer significant potential in computational pathology by enabling interpretable image analysis, automated reporting, and scalable decision support. However, their widespread clinical adoption remains limited due to the absence of reliable, automated evaluation metrics capable of identifying subtle failures such as hallucinations. To address this gap, we propose PathGLS, a novel reference-free evaluation framework that assesses pathology VLMs across three dimensions: Grounding (fine-grained visual-text alignment), Logic (entailment graph consistency using Natural Language Inference), and Stability (output variance under adversarial visual-semantic perturbations). PathGLS supports both patch-level and whole-slide image (WSI)-level analysis, yielding a comprehensive trust score. Experiments on Quilt-1M, TCGA, REG2025, PathMMU and TCGA-Sarcoma datasets demonstrate the superiority of PathGLS. Specifically, on the Quilt-1M dataset, PathGLS reveals a steep sensitivity drop of 40.2% for hallucinated reports compared to only 2.1% for BERTScore. Moreover, validation against expert-defined clinical error hierarchies reveals that PathGLS achieves a strong Spearman's rank correlation of $\rho=0.71$ ($p < 0.0001$), significantly outperforming Large Language Model (LLM)-based approaches (Gemini 3.0 Pro: $\rho=0.39$, $p < 0.0001$). These results establish PathGLS as a robust reference-free metric. By directly quantifying hallucination rates and domain shift robustness, it serves as a reliable criterion for benchmarking VLMs on private clinical datasets and informing safe deployment. Code can be found at: https://github.com/My13ad/PathGLS

cs.CV↗

arXiv2026

Aligning What EEG Can See: Structural Representations for Brain-Vision Matching

Visual decoding from electroencephalography (EEG) has emerged as a highly promising avenue for non-invasive brain-computer interfaces (BCIs). Existing EEG-based decoding methods predominantly align brain signals with the final-layer semantic embeddings of deep visual models. However, relying on these highly abstracted embeddings inevitably leads to severe cross-modal information mismatch. In this work, we introduce the concept of Neural Visibility and accordingly propose the EEG-Visible Layer Selection Strategy, aligning EEG signals with intermediate visual layers to minimize this mismatch. Furthermore, to accommodate the multi-stage nature of human visual processing, we propose a novel Hierarchically Complementary Fusion (HCF) framework that jointly integrates visual representations from different hierarchical levels. Extensive experiments demonstrate that our method achieves state-of-the-art performance, reaching an 84.6% accuracy (+21.4%) on zero-shot visual decoding on the THINGS-EEG dataset. Moreover, our method achieves up to a 129.8% performance gain across diverse EEG baselines, demonstrating its robust generalizability.

cs.CV↗

arXiv2026

Robust LLM-based Audio-Visual Speech Recognition with Sparse Modality Alignment and Visual Unit-Guided Refinement

Audio-Visual Speech Recognition (AVSR) integrates acoustic and visual information to enhance robustness in adverse acoustic conditions. Recent advances in Large Language Models (LLMs) have yielded competitive automatic speech recognition performance and shown effectiveness for AVSR. However, prior approaches project audio and visual features independently or apply shallow fusion, limiting cross-modal alignment and complementary exchange while increasing the LLM's computational load. To address this, we propose AVUR-LLM, an LLM-based Audio-Visual Speech Recognition via Sparse Modality Alignment and Visual Unit-Guided Refinement. Experiments on LRS3 demonstrate state-of-the-art results for AVSR. Under additive-noise conditions at 0 dB SNR, it achieves 37% relative improvement over the baseline system.

cs.SD↗

arXiv2026

Robustness Risk of Conversational Retrieval: Identifying and Mitigating Noise Sensitivity in Qwen3-Embedding Model

We present an empirical study of embedding-based retrieval under realistic conversational settings, where queries are short, dialogue-like, and weakly specified, and retrieval corpora contain structured conversational artifacts. Focusing on Qwen3-embedding models, we identify a deployment-relevant robustness vulnerability: under conversational retrieval without query prompting, structured dialogue-style noise can become disproportionately retrievable and intrude into top-ranked results, despite being semantically uninformative. This failure mode emerges consistently across model scales, remains largely invisible under standard clean-query benchmarks, and is significantly more pronounced in Qwen3 than in earlier Qwen variants and other widely used dense retrieval baselines. We further show that lightweight query prompting qualitatively alters retrieval behavior, effectively suppressing noise intrusion and restoring ranking stability. Our findings highlight an underexplored robustness risk in conversational retrieval and underscore the importance of evaluation protocols that reflect the complexities of deployed systems.

cs.IR↗

arXiv2025

Post-TIPS Prediction via Multimodal Interaction: A Multi-Center Dataset and Framework for Survival, Complication, and Portal Pressure Assessment

Transjugular intrahepatic portosystemic shunt (TIPS) is an established procedure for portal hypertension, but provides variable survival outcomes and frequent overt hepatic encephalopathy (OHE), indicating the necessity of accurate preoperative prognostic modeling. Current studies typically build machine learning models from preoperative CT images or clinical characteristics, but face three key challenges: (1) labor-intensive region-of-interest (ROI) annotation, (2) poor reliability and generalizability of unimodal methods, and (3) incomplete assessment from single-endpoint prediction. Moreover, the lack of publicly accessible datasets constrains research in this field. Therefore, we present MultiTIPS, the first public multi-center dataset for TIPS prognosis, and propose a novel multimodal prognostic framework based on it. The framework comprises three core modules: (1) dual-option segmentation, which integrates semi-supervised and foundation model-based pipelines to achieve robust ROI segmentation with limited annotations and facilitate subsequent feature extraction; (2) multimodal interaction, where three techniques, multi-grained radiomics attention (MGRA), progressive orthogonal disentanglement (POD), and clinically guided prognostic enhancement (CGPE), are introduced to enable cross-modal feature interaction and complementary representation integration, thus improving model accuracy and robustness; and (3) multi-task prediction, where a staged training strategy is used to perform stable optimization of survival, portal pressure gradient (PPG), and OHE prediction for comprehensive prognostic assessment. Extensive experiments on MultiTIPS demonstrate the superiority of the proposed method over state-of-the-art approaches, along with strong cross-domain generalization and interpretability, indicating its promise for clinical application. The dataset and code are available.

cs.CV↗

arXiv2025

Towards Efficient Multimodal Unified Reasoning Model via Model Merging

Although Multimodal Large Language Models (MLLMs) have demonstrated remarkable capabilities across diverse tasks, they encounter challenges in terms of reasoning efficiency, large model size and overthinking. However, existing lightweight MLLMs lack the capability to balance high efficiency and performance at a small scale. To this end, we propose Tiny-R1V, a novel lightweight 3B model that achieves faster inference and higher accuracy via a two-stage optimization, while unifying multimodal reasoning across multiple tasks with fewer inference tokens. In the first stage, Tiny-R1V introduces Length-Informed Relative Policy Optimization (LIPO), a new reinforcement learning method, to train each reasoning model, including mathematical reasoning, chart reasoning, and OCR capability. The LIPO dynamically adjusts the advantages of responses within groups by prioritizing concise yet high-quality responses to encourage the generation of shorter and more accurate responses. In the second stage, we propose Adaptive Model Merging (AMM), a training-free model merging method that merges multiple specialist models into a unified architecture. Specifically, AMM adaptively adjusts the weights of task vectors via a novel gradient projection regularization loss function, thus mitigating redundant conflicts between them. Extensive evaluations on ten widely-used reasoning benchmarks covering mathematics, structured data (charts, tables, documents), OCR, and general capabilities showcase the superior performance of Tiny-R1V, enabling lightweight models to excel in diverse multimodal reasoning tasks. Code will be available at \href{https://github.com/buptyqx/Tiny-R1V}{https://github.com/buptyqx/Tiny-R1V}

cs.AI↗

arXiv2025

AISHELL6-whisper: A Chinese Mandarin Audio-visual Whisper Speech Dataset with Speech Recognition Baselines

Whisper speech recognition is crucial not only for ensuring privacy in sensitive communications but also for providing a critical communication bridge for patients under vocal restraint and enabling discrete interaction in noise-sensitive environments. The development of Chinese mandarin audio-visual whisper speech recognition is hindered by the lack of large-scale datasets. We present AISHELL6-Whisper, a large-scale open-source audio-visual whisper speech dataset, featuring 30 hours each of whisper speech and parallel normal speech, with synchronized frontal facial videos. Moreover, we propose an audio-visual speech recognition (AVSR) baseline based on the Whisper-Flamingo framework, which integrates a parallel training strategy to align embeddings across speech types, and employs a projection layer to adapt to whisper speech's spectral properties. The model achieves a Character Error Rate (CER) of 4.13% for whisper speech and 1.11% for normal speech in the test set of our dataset, and establishes new state-of-the-art results on the wTIMIT benchmark. The dataset and the AVSR baseline codes are open-sourced at https://zutm.github.io/AISHELL6-Whisper.

eess.AS↗

arXiv2025

DiTraj: training-free trajectory control for video diffusion transformer

Diffusion Transformers (DiT)-based video generation models with 3D full attention exhibit strong generative capabilities. Trajectory control represents a user-friendly task in the field of controllable video generation. However, existing methods either require substantial training resources or are specifically designed for U-Net, do not take advantage of the superior performance of DiT. To address these issues, we propose DiTraj, a simple but effective training-free framework for trajectory control in text-to-video generation, tailored for DiT. Specifically, first, to inject the object's trajectory, we propose foreground-background separation guidance: we use the Large Language Model (LLM) to convert user-provided prompts into foreground and background prompts, which respectively guide the generation of foreground and background regions in the video. Then, we analyze 3D full attention and explore the tight correlation between inter-token attention scores and position embedding. Based on this, we propose inter-frame Spatial-Temporal Decoupled 3D-RoPE (STD-RoPE). By modifying only foreground tokens' position embedding, STD-RoPE eliminates their cross-frame spatial discrepancies, strengthening cross-frame attention among them and thus enhancing trajectory control. Additionally, we achieve 3D-aware trajectory control by regulating the density of position embedding. Extensive experiments demonstrate that our method outperforms previous methods in both video quality and trajectory controllability.

cs.CV↗

arXiv2025

FingerSplat: Contactless Fingerprint 3D Reconstruction and Generation based on 3D Gaussian Splatting

Researchers have conducted many pioneer researches on contactless fingerprints, yet the performance of contactless fingerprint recognition still lags behind contact-based methods primary due to the insufficient contactless fingerprint data with pose variations and lack of the usage of implicit 3D fingerprint representations. In this paper, we introduce a novel contactless fingerprint 3D registration, reconstruction and generation framework by integrating 3D Gaussian Splatting, with the goal of offering a new paradigm for contactless fingerprint recognition that integrates 3D fingerprint reconstruction and generation. To our knowledge, this is the first work to apply 3D Gaussian Splatting to the field of fingerprint recognition, and the first to achieve effective 3D registration and complete reconstruction of contactless fingerprints with sparse input images and without requiring camera parameters information. Experiments on 3D fingerprint registration, reconstruction, and generation prove that our method can accurately align and reconstruct 3D fingerprints from 2D images, and sequentially generates high-quality contactless fingerprints from 3D model, thus increasing the performances for contactless fingerprint recognition.

cs.CV↗

arXiv2025

MOOM: Maintenance, Organization and Optimization of Memory in Ultra-Long Role-Playing Dialogues

Memory extraction is crucial for maintaining coherent ultra-long dialogues in human-robot role-playing scenarios. However, existing methods often exhibit uncontrolled memory growth. To address this, we propose MOOM, the first dual-branch memory plugin that leverages literary theory by modeling plot development and character portrayal as core storytelling elements. Specifically, one branch summarizes plot conflicts across multiple time scales, while the other extracts the user's character profile. MOOM further integrates a forgetting mechanism, inspired by the ``competition-inhibition'' memory theory, to constrain memory capacity and mitigate uncontrolled growth. Furthermore, we present ZH-4O, a Chinese ultra-long dialogue dataset specifically designed for role-playing, featuring dialogues that average 600 turns and include manually annotated memory information. Experimental results demonstrate that MOOM outperforms all state-of-the-art memory extraction methods, requiring fewer large language model invocations while maintaining a controllable memory capacity.

cs.CL↗

arXiv2025

A Single-step Accurate Fingerprint Registration Method Based on Local Feature Matching

Distortion of the fingerprint images leads to a decline in fingerprint recognition performance, and fingerprint registration can mitigate this distortion issue by accurately aligning two fingerprint images. Currently, fingerprint registration methods often consist of two steps: an initial registration based on minutiae, and a dense registration based on matching points. However, when the quality of fingerprint image is low, the number of detected minutiae is reduced, leading to frequent failures in the initial registration, which ultimately causes the entire fingerprint registration process to fail. In this study, we propose an end-to-end single-step fingerprint registration algorithm that aligns two fingerprints by directly predicting the semi-dense matching points correspondences between two fingerprints. Thus, our method minimizes the risk of minutiae registration failure and also leverages global-local attentions to achieve end-to-end pixel-level alignment between the two fingerprints. Experiment results prove that our method can achieve the state-of-the-art matching performance with only single-step registration, and it can also be used in conjunction with dense registration algorithms for further performance improvements.

cs.CV↗

arXiv2025

Automatic Synthesis of High-Quality Triplet Data for Composed Image Retrieval

As a challenging vision-language (VL) task, Composed Image Retrieval (CIR) aims to retrieve target images using multimodal (image+text) queries. Although many existing CIR methods have attained promising performance, their reliance on costly, manually labeled triplets hinders scalability and zero-shot capability. To address this issue, we propose a scalable pipeline for automatic triplet generation, along with a fully synthetic dataset named Composed Image Retrieval on High-quality Synthetic Triplets (CIRHS). Our pipeline leverages a large language model (LLM) to generate diverse prompts, controlling a text-to-image generative model to produce image pairs with identical elements in each pair, which are then filtered and reorganized to form the CIRHS dataset. In addition, we introduce Hybrid Contextual Alignment (CoAlign), a novel CIR framework, which can accomplish global alignment and local reasoning within a broader context, enabling the model to learn more robust and informative representations. By utilizing the synthetic CIRHS dataset, CoAlign achieves outstanding zero-shot performance on three commonly used benchmarks, demonstrating for the first time the feasibility of training CIR models on a fully synthetic dataset. Furthermore, under supervised training, our method outperforms all the state-of-the-art supervised CIR approaches, validating the effectiveness of our proposed retrieval framework. The code and the CIRHS dataset will be released soon.

cs.CV↗

arXiv2025

Multi-Channel Sequence-to-Sequence Neural Diarization: Experimental Results for The MISP 2025 Challenge

This paper describes the speaker diarization system developed for the Multimodal Information-Based Speech Processing (MISP) 2025 Challenge. First, we utilize the Sequence-to-Sequence Neural Diarization (S2SND) framework to generate initial predictions using single-channel audio. Then, we extend the original S2SND framework to create a new version, Multi-Channel Sequence-to-Sequence Neural Diarization (MC-S2SND), which refines the initial results using multi-channel audio. The final system achieves a diarization error rate (DER) of 8.09% on the evaluation set of the competition database, ranking first place in the speaker diarization task of the MISP 2025 Challenge.

eess.AS↗

arXiv2025

R1-ShareVL: Incentivizing Reasoning Capability of Multimodal Large Language Models via Share-GRPO

In this work, we aim to incentivize the reasoning ability of Multimodal Large Language Models (MLLMs) via reinforcement learning (RL) and develop an effective approach that mitigates the sparse reward and advantage vanishing issues during RL. To this end, we propose Share-GRPO, a novel RL approach that tackle these issues by exploring and sharing diverse reasoning trajectories over expanded question space. Specifically, Share-GRPO first expands the question space for a given question via data transformation techniques, and then encourages MLLM to effectively explore diverse reasoning trajectories over the expanded question space and shares the discovered reasoning trajectories across the expanded questions during RL. In addition, Share-GRPO also shares reward information during advantage computation, which estimates solution advantages hierarchically across and within question variants, allowing more accurate estimation of relative advantages and improving the stability of policy training. Extensive evaluations over six widely-used reasoning benchmarks showcase the superior performance of our method. Code will be available at https://github.com/HJYao00/R1-ShareVL.

cs.CV↗

arXiv2025

OLMA: One Loss for More Accurate Time Series Forecasting

Time series forecasting faces two important but often overlooked challenges. Firstly, the inherent random noise in the time series labels sets a theoretical lower bound for the forecasting error, which is positively correlated with the entropy of the labels. Secondly, neural networks exhibit a frequency bias when modeling the state-space of time series, that is, the model performs well in learning certain frequency bands but poorly in others, thus restricting the overall forecasting performance. To address the first challenge, we prove a theorem that there exists a unitary transformation that can reduce the marginal entropy of multiple correlated Gaussian processes, thereby providing guidance for reducing the lower bound of forecasting error. Furthermore, experiments confirm that Discrete Fourier Transform (DFT) can reduce the entropy in the majority of scenarios. Correspondingly, to alleviate the frequency bias, we jointly introduce supervision in the frequency domain along the temporal dimension through DFT and Discrete Wavelet Transform (DWT). This supervision-side strategy is highly general and can be seamlessly integrated into any supervised learning method. Moreover, we propose a novel loss function named OLMA, which utilizes the frequency domain transformation across both channel and temporal dimensions to enhance forecasting. Finally, the experimental results on multiple datasets demonstrate the effectiveness of OLMA in addressing the above two challenges and the resulting improvement in forecasting accuracy. The results also indicate that the perspectives of entropy and frequency bias provide a new and feasible research direction for time series forecasting. The code is available at: https://github.com/Yuyun1011/OLMA-One-Loss-for-More-Accurate-Time-Series-Forecasting.

cs.LG↗