Large Vision Models (LVMs) have demonstrated impressive capabilities by leveraging large-scale generative pre-training on visual sequences. However, training these models on massive datasets of single images and image sequences can be computationally expensive and limit their accessibility to researchers without substantial computational resources. This paper introduces LVM-Lite, a new two-stage learning pipeline for more efficient and effective LVM training. In the first stage, the model is pre-trained on a large corpus of single images. Subsequently, in the second stage, the model is fine-tuned on curated long image/video sequences. This decoupled training approach substantially accelerates the training process, achieving up to 2.7× speedup compared to baseline LVM training. Extensive experiments demonstrate that LVM-Lite achieves competitive performance on various generative and discriminative benchmarks while maintaining high training efficiency and strong scalability. https://github.com/UCSC-VLAA/LVM-Lite

The precise assessment of major depressive disorder (MDD) is of vital importance for clinical management. Electroencephalographic (EEG) is expected to provide an objective neurophysiological alternative for this purpose, such as the abnormal functional connectivity characteristics that are closely related to MDD. However, the current approaches using a single functional connectivity matrix cannot fully capture the information in EEG signals, and combining multiple matrices would introduces redundancy. This study proposes a Multi-head and Channel Twin Self-Attention Network (MCTSA-Net), a diagnosis model based on multi-feature fusion and self-attention mechanisms. Based on coupling methods and graph theory, it extracts different functional connectivity matrices and topological attributes from EEG signals, and the best combinations and fusion of these properties are selected, including Pearson correlation coefficient (PCC), phase locking value (PLV), strength feature (sf), and clustering coefficient (cc). Then, multi-head and channel selfattention mechanisms are incorporated to capture complex relationships between topological properties and frequency bands, automatically assigning appropriate weights for classification. The model achieves an accuracy of 98.53 % at the MODMA dataset, outperforming existing MDD diagnostic methods. Additionally, the SHAP analysis shows that the gamma band contributes the most significantly to the classification outcomes, further exploring the interpretability of the diagnosis framework.

Magnetic resonance imaging (MRI) is essential for studying neurological and psychiatric disorders, but its analysis is limited by sparse connectivity and individual variability. Most existing methods rely on pairwise Pearson correlation, which captures only direct interactions and overlooks indirect, mediated connections, leading to oversimplified brain networks. To address this, we propose Dynamic Graph Enhancement -Multi-level Attention Weighting (DGE-MAW), a data-centered approach that improves structural representation and diagnostic interpretability. It comprises two main modules: Dynamic Graph Enhancement dynamically updates the graph structure to identify indirect connections. By integrating intermediate nodes, nonlinear activations, and topological indicators, it mimics neural transmission pathways and reconstructs more informative brain networks, reducing information loss. Multi-level Attention Weighting introduces a hierarchical attention mechanism inspired by biological signal transmission, capturing individual, regional, and global semantics. This enables interpretable and adaptive integration of multimodal features. Experiments show that DGE-MAW significantly enhances data quality, interpretability, multimodal fusion, and diagnostic accuracy. On the ABIDE and ADHD-200 datasets, it outperforms state-of-the-art model-centric methods across various neuroimaging tasks. By modeling indirect connections and incorporating prior brain knowledge, it demonstrates the potential of MRI augmentation. Code and resources are available at: https://github.com/A3-247/DGE-MAW.

Functional connectivity changes detected via functional magnetic resonance imaging (fMRI) often precede clinical symptoms of neurological and psychiatric disorders, making them valuable biomarkers for early diagnosis and monitoring. This study introduces a multi-modal fusion framework that combines fMRI-derived connectivity with demographic data, using a multi-task learning strategy informed by medical domain knowledge. We propose a novel HemiSynap Polarity Attention mechanism to enhance feature extraction by modeling inter-hemispheric interactions and capturing polarityspecific (positive/negative) connectivity patterns. An adversarial multi-task training scheme aligns classification and regression objectives, promoting loss sharing and effective information fusion. The model was validated on three public datasets (ABIDE, ADHD-200, ADNI), showing notable improvements over state-of-the-art methods. Furthermore, we conducted statistical analyses of attention patterns across brain networks and regions, demonstrating the model's interpretability and offering insights into its clinical relevance for aiding diagnosis. The code is available at: https://github.com/A3-247/HPA-AT

The machine learning community has witnessed a drastic change in the training pipeline, pivoted by those “foundation models” with unprecedented scales. However, the field of adversarial training is lagging behind, predominantly centered around small model sizes like ResNet-50, and tiny and low-resolution datasets like CIFAR-10. To bridge this transformation gap, this paper provides a modern re-examination with adversarial training, investigating its potential benefits when applied at scale. Additionally, we introduce an efficient and effective training strategy to enable adversarial training with giant models and web-scale data at an affordable computing cost. We denote this newly introduced framework as AdvXL. Empirical results demonstrate that AdvXL establishes new state-of-the-art robust accuracy records under AutoAttack on ImageNet-1K. For example, by training on DataComp-1B dataset, our AdvXL empowers a vanilla ViT-g model to substantially surpass the previous records of l∞-, l2-, and l1-robust accuracy by margins of 11.4%, 14.2% and 12.9%, respectively. This achievement posits AdvXL as a pioneering approach, charting a new trajectory for the efficient training of robust visual representations at significantly larger scales. Our code is available at https://github.com/UCSC-VLAA/AdvXL.

The Segment Anything Model (SAM) has garnered significant attention for its versatile segmentation abilities and intuitive prompt-based interface. However, its application in medical imaging presents challenges, requiring either substantial training costs and extensive medical datasets for full model fine-tuning or high-quality prompts for optimal performance. This paper introduces H-SAM: a prompt-free adaptation of SAM tailored for efficient fine-tuning of medical images via a two-stage hierarchical decoding procedure. In the initial stage, H-SAM employs SAM's original decoder to generate a prior probabilistic mask, guiding a more intricate decoding process in the second stage. Specifically, we propose two key designs: 1) A class-balanced, mask-guided self-attention mechanism addressing the unbalanced label distribution, enhancing image embedding; 2) A learnable mask cross-attention mechanism spatially modulating the interplay among different image regions based on the prior mask. Moreover, the inclusion of a hierarchical pixel decoder in H-SAM enhances its proficiency in capturing fine-grained and localized details. This approach enables SAM to effectively integrate learned medical priors, facilitating enhanced adaptation for medical image segmentation with limited samples. Our H-SAM demonstrates a 4.78% improvement in average Dice compared to existing prompt-free SAM variants for multi-organ segmentation using only 10% of 2D slices. Notably, without using any unlabeled data, H-SAM even outperforms state-of-the-art semisupervised models relying on extensive unlabeled training data across various medical datasets. Our code is available at https://github.com/Cccccczh404/H-SAM.

Volumetric optical microscopy using non-diffracting beams enables rapid imaging of 3D volumes by projecting them axially to 2D images but lacks crucial depth information. Addressing this, we introduce MicroDiffusion, a pi-oneering tool facilitating high-quality, depth-resolved 3D volume reconstruction from limited 2D projections. While existing Implicit Neural Representation (INR) models often yield incomplete outputs and Denoising Diffusion Prob-abilistic Models (DDPM) excel at capturing details, our method integrates INR's structural coherence with DDPM's fine-detail enhancement capabilities. We pretrain an INR model to transform 2D axially-projected images into a pre-liminary 3D volume. This pretrained INR acts as a global prior guiding DDPM's generative process through a linear interpolation between INR outputs and noise inputs. This strategy enriches the diffusion process with structured 3D information, enhancing detail and reducing noise in localized 2D images. By conditioning the diffusion model on the closest 2D projection, MicroDiffusion substantially enhances fidelity in resulting 3D reconstructions, surpassing INR and standard DDPM outputs with unparalleled image quality and structural fidelity. Our code and dataset are available at https://github.com/UCSC-VLAA/MicroDiffusion.

This paper enhances image-GPT (iGPT), one of the pioneering works that introduce autoregressive pretraining to predict the next pixels for visual representation learning. Two simple yet essential changes are made. First, we shift the prediction target from raw pixels to semantic tokens, enabling a higher-level understanding of visual content. Second, we supplement the autoregressive modeling by instructing the model to predict not only the next tokens but also the visible tokens. This pipeline is particularly effective when semantic tokens are encoded by discriminatively trained models, such as CLIP. We introduce this novel approach as D-iGPT. Extensive experiments showcase that D-iGPT excels as a strong learner of visual representations: A notable achievement is its compelling performance on the ImageNet-1K dataset — by training on publicly available datasets, D-iGPT unprecedentedly achieves 90.0% top-1 accuracy with a vanilla ViT-H. Additionally, D-iGPT shows strong generalization on the downstream task. Code is available at https://github.com/OliverRensu/D-iGPT.

Accurate electroencephalogram (EEG) pattern decoding for specific mental tasks is one of the key steps for the development of brain-computer interface (BCI), which is quite challenging due to the considerably low signal-to-noise ratio of EEG collected at the brain scalp. Machine learning provides a promising technique to optimize EEG patterns toward better decoding accuracy. However, existing algorithms do not effectively explore the underlying data structure capturing the true EEG sample distribution and, hence, can only yield a suboptimal decoding accuracy. To uncover the intrinsic distribution structure of EEG data, we propose a clustering-based multitask feature learning algorithm for improved EEG pattern decoding. Specifically, we perform affinity propagation-based clustering to explore the subclasses (i.e., clusters) in each of the original classes and then assign each subclass a unique label based on a one-versus-all encoding strategy. With the encoded label matrix, we devise a novel multitask learning algorithm by exploiting the subclass relationship to jointly optimize the EEG pattern features from the uncovered subclasses. We then train a linear support vector machine with the optimized features for EEG pattern decoding. Extensive experimental studies are conducted on three EEG data sets to validate the effectiveness of our algorithm in comparison with other state-of-the-art approaches. The improved experimental results demonstrate the outstanding superiority of our algorithm, suggesting its prominent performance for EEG pattern decoding in BCI applications.