Fengling Li

Chenyv Liu, Wentao Tan, Lei Zhu et al.

Standard vision-language-action (VLA) models rely on fitting statistical data priors, limiting their robust understanding of underlying physical dynamics. Reinforcement learning enhances physical grounding through exploration yet typically relies on external reward signals that remain isolated from the agent's internal states. World action models have emerged as a promising paradigm that integrates imagination and control to enable predictive planning. However, they rely on implicit context modeling, lacking explicit mechanisms for self-improvement. To solve these problems, we propose Self-Correcting VLA (SC-VLA), which achieve self-improvement by intrinsically guiding action refinement through sparse imagination. We first design sparse world imagination by integrating auxiliary predictive heads to forecast current task progress and future trajectory trends, thereby constraining the policy to encode short-term physical evolution. Then we introduce the online action refinement module to reshape progress-dependent dense rewards, adjusting trajectory orientation based on the predicted sparse future states. Evaluations on challenging robot manipulation tasks from simulation benchmarks and real-world settings demonstrate that SC-VLA achieve state-of-the-art performance, yielding the highest task throughput with 16% fewer steps and a 9% higher success rate than the best-performing baselines, alongside a 14% gain in real-world experiments. Code is available at https://github.com/Kisaragi0/SC-VLA.

Yipeng Chen, Wentao Tan, Lei Zhu et al.

Existing Vision-Language-Action (VLA) models often struggle to generalize to long-horizon tasks due to their heavy reliance on immediate observations. While recent studies incorporate retrieval mechanisms or extend context windows to handle procedural tasks, they often struggle to capture Non-Markovian dependencies, where optimal actions rely solely on specific past states rather than the current observation. To address this, we introduce Keyframe-Chaining VLA, a framework that extracts and links key historical frames to model long-horizon dependencies. Specifically, we propose an automatic keyframe selector that learns a discriminative embedding space, effectively identifying distinct state transitions. To capture task-critical information, we design a progress-aware query mechanism that dynamically retrieves historical frames based on their temporal relevance to the current execution phase. These selected keyframes are integrated into the VLA as interleaved visual tokens, explicitly grounding the policy in the long-horizon temporal context. Finally, we introduce a suite of four Non-Markovian manipulation tasks built upon the ManiSkill simulator to measure task success rates. Experimental results demonstrate that our method achieves superior performance, effectively tackling robot manipulation tasks characterized by long-horizon temporal dependencies. Code is available at https://github.com/cytoplastm/KC-VLA.

Yongquan Yang, Fengling Li, Yani Wei et al.

In breast cancer, neoadjuvant chemotherapy (NAC) provides a standard treatment option for patients who have locally advanced cancer and some large operable tumors. A patient will have better prognosis when he has achieved a pathological complete response (pCR) with the treatment of NAC. There has been a trend to directly predict pCR to NAC from histological images based on deep learning (DL). However, the DL-based predictive models numerically have better performances in internal validation than in external validation. In this paper, we aim to alleviate this situation with an intrinsic approach. We propose an experts' cognition-driven ensemble deep learning (ECDEDL) approach. Taking the cognition of both pathology and artificial intelligence experts into consideration to improve the generalization of the predictive model to the external validation, ECDEDL can intrinsically approximate the working paradigm of a human being which will refer to his various working experiences to make decisions. ECDEDL was validated with 695 WSIs collected from the same center as the primary dataset to develop the predictive model and perform the internal validation, and was also validated with 340 WSIs collected from other three centers as the external dataset to perform the external validation. In external validation, ECDEDL improves the AUCs of pCR prediction from 61.52(59.80-63.26) to 67.75(66.74-68.80) and the Accuracies of pCR prediction from 56.09(49.39-62.79) to 71.01(69.44-72.58). ECDEDL was quite effective for external validation of predicting pCR to NAC from histological images in breast cancer, numerically approximating the internal validation.

Siyuan Xu, Tianshi Wang, Fengling Li et al.

Vision-Language-Action models (VLAs) have demonstrated strong potential for embodied AI, yet their deployment on resource-limited robots remains challenging due to high memory and computational demands. While Post-Training Quantization (PTQ) provides an efficient solution, directly applying PTQ to VLAs often results in severe performance degradation during sequential control. We identify temporal error accumulation as a key factor, where quantization perturbations at the vision-language-to-action interface are progressively amplified, leading to kinematic drift in executed trajectories. To address this issue, we propose Drift-Aware Post-Training Quantization (DA-PTQ), which formulates quantization as a drift-aware optimization problem over sequential decision processes. DA-PTQ consists of two components: (1) Cross-Space Representation Compensation, which mitigates structured distortions between multimodal representations and action space to improve action consistency, and (2) Motion-Driven Mixed-Precision Allocation, which assigns bit-widths by minimizing trajectory-level motion errors. Extensive experiments show that DA-PTQ significantly reduces kinematic drift and achieves comparable performance to full-precision models under low-bit settings, enabling practical deployment of VLAs on resource-limited robotic platforms.

Xinyao Li, Jingjing Li, Fengling Li et al.

Efficiently utilizing rich knowledge in pretrained models has become a critical topic in the era of large models. This work focuses on adaptively utilizing knowledge from multiple source-pretrained models to an unlabeled target domain without accessing the source data. Despite being a practically useful setting, existing methods require extensive parameter tuning over each source model, which is computationally expensive when facing abundant source domains or larger source models. To address this challenge, we propose a novel approach which is free of the parameter tuning over source backbones. Our technical contribution lies in the Bi-level ATtention ENsemble (Bi-ATEN) module, which learns both intra-domain weights and inter-domain ensemble weights to achieve a fine balance between instance specificity and domain consistency. By slightly tuning source bottlenecks, we achieve comparable or even superior performance on a challenging benchmark DomainNet with less than 3% trained parameters and 8 times of throughput compared with SOTA method. Furthermore, with minor modifications, the proposed module can be easily equipped to existing methods and gain more than 4% performance boost. Code is available at https://github.com/TL-UESTC/Bi-ATEN.

Xinyao Li, Yuke Li, Zhekai Du et al.

Large vision-language models (VLMs) like CLIP have demonstrated good zero-shot learning performance in the unsupervised domain adaptation task. Yet, most transfer approaches for VLMs focus on either the language or visual branches, overlooking the nuanced interplay between both modalities. In this work, we introduce a Unified Modality Separation (UniMoS) framework for unsupervised domain adaptation. Leveraging insights from modality gap studies, we craft a nimble modality separation network that distinctly disentangles CLIP's features into language-associated and vision-associated components. Our proposed Modality-Ensemble Training (MET) method fosters the exchange of modality-agnostic information while maintaining modality-specific nuances. We align features across domains using a modality discriminator. Comprehensive evaluations on three benchmarks reveal our approach sets a new state-of-the-art with minimal computational costs. Code: https://github.com/TL-UESTC/UniMoS

Xinyao Li, Yinjie Min, Hongbo Chen et al.

Fine-tuning large pretrained vision-language models (VLMs) has emerged as a prevalent paradigm for downstream adaptation, yet it faces a critical trade-off between domain specificity and domain generalization (DG) ability. Current methods typically fine-tune a universal model on the entire dataset, which potentially compromises the ability to generalize to unseen domains. To fill this gap, we provide a theoretical understanding of the generalization ability for VLM fine-tuning, which reveals that training multiple parameter-efficient expert models on partitioned source domains leads to better generalization than fine-tuning a universal model. Inspired by this finding, we propose a two-step domain-expert-Guided DG (GuiDG) framework. GuiDG first employs prompt tuning to obtain source domain experts, then introduces a Cross-Modal Attention module to guide the fine-tuning of the vision encoder via adaptive expert integration. To better evaluate few-shot DG, we construct ImageNet-DG from ImageNet and its variants. Extensive experiments on standard DG benchmarks and ImageNet-DG demonstrate that GuiDG improves upon state-of-the-art fine-tuning methods while maintaining efficiency.

Peng-Fei Zhang, Ying Cheng, Xiaofan Sun et al.

Autonomous agents are increasingly expected to operate in complex, dynamic, and uncertain environments, performing tasks such as manipulation, navigation, and decision-making. Achieving these capabilities requires agents to understand the underlying mechanisms and dynamics of the world, moving beyond reactive control or simple replication of observed states. This motivates the development of world models as internal representations that encode environmental states, capture dynamics, and support prediction, planning, and reasoning. Despite growing interest, the definition, scope, architectures, and essential capabilities of world models remain ambiguous. In this survey, we go beyond prescribing a fixed definition and limiting our scope to methods explicitly labeled as world models. Instead, we examine approaches that exhibit the core capabilities of world models through a review of methods in robotic manipulation. We analyze their roles across perception, prediction, and control, identify key challenges and solutions, and distill the core components, capabilities, and functions that a fully realized world model should possess. Building on this analysis, we aim to motivate further development toward generalizable and practical world models for robotics.

Pengyu Liu, Guohua Dong, Dan Guo et al.

In daily life, we encounter diverse external stimuli, such as images, sounds, and videos. As research in multimodal stimuli and neuroscience advances, fMRI-based brain decoding has become a key tool for understanding brain perception and its complex cognitive processes. Decoding brain signals to reconstruct stimuli not only reveals intricate neural mechanisms but also drives progress in AI, disease treatment, and brain-computer interfaces. Recent advancements in neuroimaging and image generation models have significantly improved fMRI-based decoding. While fMRI offers high spatial resolution for precise brain activity mapping, its low temporal resolution and signal noise pose challenges. Meanwhile, techniques like GANs, VAEs, and Diffusion Models have enhanced reconstructed image quality, and multimodal pre-trained models have boosted cross-modal decoding tasks. This survey systematically reviews recent progress in fMRI-based brain decoding, focusing on stimulus reconstruction from passive brain signals. It summarizes datasets, relevant brain regions, and categorizes existing methods by model structure. Additionally, it evaluates model performance and discusses their effectiveness. Finally, it identifies key challenges and proposes future research directions, offering valuable insights for the field. For more information and resources related to this survey, visit https://github.com/LpyNow/BrainDecodingImage.

Zhekai Du, Xinyao Li, Fengling Li et al.

Conventional Unsupervised Domain Adaptation (UDA) strives to minimize distribution discrepancy between domains, which neglects to harness rich semantics from data and struggles to handle complex domain shifts. A promising technique is to leverage the knowledge of large-scale pre-trained vision-language models for more guided adaptation. Despite some endeavors, current methods often learn textual prompts to embed domain semantics for source and target domains separately and perform classification within each domain, limiting cross-domain knowledge transfer. Moreover, prompting only the language branch lacks flexibility to adapt both modalities dynamically. To bridge this gap, we propose Domain-Agnostic Mutual Prompting (DAMP) to exploit domain-invariant semantics by mutually aligning visual and textual embeddings. Specifically, the image contextual information is utilized to prompt the language branch in a domain-agnostic and instance-conditioned way. Meanwhile, visual prompts are imposed based on the domain-agnostic textual prompt to elicit domain-invariant visual embeddings. These two branches of prompts are learned mutually with a cross-attention module and regularized with a semantic-consistency loss and an instance-discrimination contrastive loss. Experiments on three UDA benchmarks demonstrate the superiority of DAMP over state-of-the-art approaches.

Xinyao Li, Jingjing Li, Fengling Li et al.

Recently, vision-language pretraining has emerged as a transformative technique that integrates the strengths of both visual and textual modalities, resulting in powerful vision-language models (VLMs). Leveraging web-scale pretraining data, these models exhibit strong zero-shot capabilities. However, their performance often deteriorates when confronted with domain-specific or specialized generalization tasks. To address this, a growing body of research focuses on transferring or generalizing the rich knowledge embedded in VLMs to various downstream applications. This survey aims to comprehensively summarize the generalization settings, methodologies, benchmarking and results in VLM literatures. Delving into the typical VLM structures, current literatures are categorized into prompt-based, parameter-based and feature-based methods according to the transferred modules. The differences and characteristics in each category are furthered summarized and discussed by revisiting the typical transfer learning (TL) settings, providing novel interpretations for TL in the era of VLMs. Popular benchmarks for VLM generalization are further introduced with thorough performance comparisons among the reviewed methods. Following the advances in large-scale generalizable pretraining, this survey also discusses the relations and differences between VLMs and up-to-date multimodal large language models (MLLM), e.g., DeepSeek-VL. By systematically reviewing the surging literatures in vision-language research from a novel and practical generalization prospective, this survey contributes to a clear landscape of current and future multimodal researches.

Wencheng Ye, Tianshi Wang, Lei Zhu et al.

Recent Vision-Language-Action (VLA) models have shown impressive flexibility and generalization, yet their deployment in robotic manipulation remains limited by heavy computational overhead and inference latency. In this work, we present ActDistill, a general action-guided self-derived distillation framework that transfers the action prediction capability of any existing VLA model to a lightweight counterpart. Unlike previous efficiency strategies that primarily emphasize vision-language correlations, ActDistill leverages action priors to guide knowledge transfer and model compression, achieving action-oriented efficiency for VLA models. Specifically, we employ a well-trained VLA model as the teacher and introduce a graph-structured encapsulation strategy to explicitly model the hierarchical evolution of action prediction. The student model, derived from the graph-encapsulated teacher, is further equipped with a dynamic router that adaptively selects computation paths based on action prediction demands, guided by hierarchical graph-informed supervision to ensure smooth and efficient evolution. During inference, graph-related auxiliary components are removed, allowing the student to execute only dynamically routed layers and predict high-precision actions with minimal computation and latency. Experiments on embodied benchmarks demonstrate that ActDistill achieves comparable or superior performance to full-scale VLA models while reducing computation by over 50% with up to 1.67 times speedup, thereby establishing a general paradigm toward efficient embodied intelligence.

Yongquan Yang, Fengling Li, Yani Wei et al.

Recent studies have demonstrated the effectiveness of the combination of machine learning and logical reasoning, including data-driven logical reasoning, knowledge driven machine learning and abductive learning, in inventing advanced technologies for different artificial intelligence applications. One-step abductive multi-target learning (OSAMTL), an approach inspired by abductive learning, via simply combining machine learning and logical reasoning in a one-step balanced multi-target learning way, has as well shown its effectiveness in handling complex noisy labels of a single noisy sample in medical histopathology whole slide image analysis (MHWSIA). However, OSAMTL is not suitable for the situation where diverse noisy samples (DiNS) are provided for a learning task. In this paper, giving definition of DiNS, we propose one-step abductive multi-target learning with DiNS (OSAMTL-DiNS) to expand the original OSAMTL to handle complex noisy labels of DiNS. Applying OSAMTL-DiNS to tumour segmentation for breast cancer in MHWSIA, we show that OSAMTL-DiNS is able to enable various state-of-the-art approaches for learning from noisy labels to achieve more rational predictions. We released a model pre-trained with OSAMTL-DiNS for tumour segmentation in HE-stained pre-treatment biopsy images in breast cancer, which has been successfully applied as a pre-processing tool to extract tumour-associated stroma compartment for predicting the pathological complete response to neoadjuvant chemotherapy in breast cancer.

Fengling Li, Tong Wang, Lei Zhu et al.

Supervised cross-modal hashing aims to embed the semantic correlations of heterogeneous modality data into the binary hash codes with discriminative semantic labels. Because of its advantages on retrieval and storage efficiency, it is widely used for solving efficient cross-modal retrieval. However, existing researches equally handle the different tasks of cross-modal retrieval, and simply learn the same couple of hash functions in a symmetric way for them. Under such circumstance, the uniqueness of different cross-modal retrieval tasks are ignored and sub-optimal performance may be brought. Motivated by this, we present a Task-adaptive Asymmetric Deep Cross-modal Hashing (TA-ADCMH) method in this paper. It can learn task-adaptive hash functions for two sub-retrieval tasks via simultaneous modality representation and asymmetric hash learning. Unlike previous cross-modal hashing approaches, our learning framework jointly optimizes semantic preserving that transforms deep features of multimedia data into binary hash codes, and the semantic regression which directly regresses query modality representation to explicit label. With our model, the binary codes can effectively preserve semantic correlations across different modalities, meanwhile, adaptively capture the query semantics. The superiority of TA-ADCMH is proved on two standard datasets from many aspects.