Vision-Language Models (VLMs) have shown significant potential in surgical scene analysis, yet existing models are limited by frame-level datasets and lack high-quality video data with procedural surgical knowledge. To address these challenges, we make the following contributions: (i) SurgPub-Video, a comprehensive dataset of over 3,000 surgical videos and 25 million annotated frames across 11 specialities, sourced from peer-reviewed clinical journals, (ii) SurgLLaVA-Video, a specialized VLM for surgical video understanding, built upon the TinyLLaVA-Video architecture that supports both video-level and frame-level inputs, and (iii) a video-level surgical Visual Question Answering (VQA) benchmark, covering diverse 11 surgical specialities, such as vascular, cardiology, and thoracic. Extensive experiments, conducted on the proposed benchmark and three additional surgical downstream tasks (action recognition, skill assessment, and triplet recognition), show that SurgLLaVA-Video significantly outperforms both general-purpose and surgical-specific VLMs with only three billion parameters.

Continual Learning (CL) is recognized to be a storage-efficient and privacy-protecting approach for learning from sequentially-arriving medical sites. However, most existing CL methods assume that each site is fully labeled, which is impractical due to budget and expertise constraint. This paper studies the Semi-Supervised Continual Learning (SSCL) that adopts partially-labeled sites arriving over time, with each site delivering only limited labeled data while the majority remains unlabeled. In this regard, it is challenging to effectively utilize unlabeled data under dynamic cross-site domain gaps, leading to intractable model forgetting on such unlabeled data. To address this problem, we introduce a novel Disentanglement-guided Consolidation and Consistency Training (DC2T) framework, which roots in an Online Semi-Supervised representation Disentanglement (OSSD) perspective to excavate content representations of partially labeled data from sites arriving over time. Moreover, these content representations are required to be consolidated for site-invariance and calibrated for style-robustness, in order to alleviate forgetting even in the absence of ground truth. Specifically, for the invariance on previous sites, we retain historical content representations when learning on a new site, via a Content-inspired Parameter Consolidation (CPC) method that prevents altering the model parameters crucial for content preservation. For the robustness against style variation, we develop a Style-induced Consistency Training (SCT) scheme that enforces segmentation consistency over style-related perturbations to recalibrate content encoding. We extensively evaluate our method on fundus and cardiac image segmentation, indicating the advantage over existing SSCL methods for alleviating forgetting on unlabeled data.

Abnormal traffic detection is the core component of the network intrusion detection system. Although semisupervised methods can detect zero-day attack traffic, previous work suffers from high false alarms because the trained model is simply based on normal traffic. In this article, we propose an accurate abnormal traffic detection method using pseudoanomaly, consisting of an efficient feature extraction framework and a novel denoise autoencoder-generative adversarial network (DAE-GAN) model. The feature extraction framework adopts an innovative packet window scheme to extract spatial and temporal features from traffic flows. The DAE-GAN model has multiple DAEs to achieve efficient data augmentation and generate high-quality pseudoanomalies. The pseudoanomalies are obtained by adding noise on normal traffic and enhanced by adversarial learning in DAE-GAN. Our semisupervised detection method, exploiting both normal data and generated pseudoanomalies, achieves a precision of 98.6% on the NSL-KDD dataset and 98.5% on the UNSW-NB15 dataset. Compared with the state-of-the-art, the detection precision and recall under different user behaviors are significantly improved. The evaluation on four attack datasets shows that our method has a high flow-wise precision of over 99% and a high recall of 60.6%.