Industrial anomaly detection for 2D objects has gained significant attention and achieved progress in anomaly detection (AD) methods. However, identifying 3D depth anomalies using only 2D information is insufficient. Despite explicitly fusing depth information into RGB images or using point cloud backbone networks to extract depth features, both approaches struggle to adequately represent 3D information in multimodal scenarios due to the disparities among different modal information. Additionally, due to the scarcity of abnormal samples in industrial data, especially in multimodal scenarios, it is necessary to perform anomaly generation to simulate real-world abnormal samples. Therefore, we propose a novel unified multimodal anomaly detection framework to address these issues. Our contributions consist of 3 key aspects. (1) We extract visible depth information from 3D point cloud data simply and use 2D RGB images to represent appearance, which disentangles depth and appearance to support unified anomaly generation. (2) Benefiting from the flexible input representation, the proposed Multi-Scale Gaussian Anomaly Generator and Unified Texture Anomaly Generator can generate richer anomalies in RGB and depth. (3) All modules share parameters for both RGB and depth data, effectively bridging 2D and 3D anomaly detection. Subsequent modules can directly leverage features from both modalities without complex fusion. Experiments show our method outperforms state-of-the-art (SOTA) on MVTec-3D AD and Eyecandies datasets. Code available at: https://github.com/Xantastic/BridgeNet

Epilepsy as the most common neurological disorder globally has drawn more and more attention. However, it is time-consuming and labor-intensive for manual detection of interictal epileptic discharges (IEDs). Thus, there is an urgent need to develop an efficient and automated detection approach as a more accurate diagnostic alternative for epilepsy detection. Recently, fuzzy broad learning system (FBLS) has been recognized as an alternative to deep learning and utilized in various fields. Nevertheless, FBLS ignores the locally invariant property of data. To effectively address this issue and further improve the performance of FBLS, a novel graph-regularized fuzzy broad learning system (GFBLS) is first proposed based on graph regularization . Moreover, an automated GFBLS-based approach is proposed for IEDs detection from EEG recordings. In the proposed method, graph convolutional neural networks (GCN) is firstly utilized to extract features from line graphs with undirected connections, which are constructed by EEG recordings, then extracted features by GCN are fed into GFBLS for IEDs detection. The experimental results demonstrated that GFBLS can achieve accuracy of 92.20%, specificity of 90.90% and precision of 91.13% with the training time of only 31.6 s, which is superior or comparable performance compared with other state-of-the-art approaches.