TFD-Net: Transformer Deviation Network for Weakly Supervised Anomaly Detection
Hongping Gan,Hejie Zheng,Zhangfa Wu,Chunyan Ma,Jie Liu
DOI: https://doi.org/10.1109/tnsm.2024.3485545
2024-01-01
IEEE Transactions on Network and Service Management
Abstract:Deep Learning (DL)-based weakly supervised anomaly detection methods enhance the security and performance of communication and networks by promptly identifying and addressing anomalies within imbalanced samples, thus ensuring reliable communication and smooth network operations. However, existing DL-based methods often overly emphasize the local feature representations of samples, thereby neglecting the long-range dependencies and the prior knowledge of the samples, which imposes potential limitations on anomaly detection with a limited number of abnormal samples. To mitigate these challenges, we propose a Transformer deviation network for weakly supervised anomaly detection, called TFD-Net, which can effectively leverage the interdependencies and data priors of samples, yielding enhanced anomaly detection performance. Specifically, we first use a Transformer-based feature extraction module that proficiently captures the dependencies of global features in the samples. Subsequently, TFD-Net employs an anomaly score generation module to obtain corresponding anomaly scores. Finally, we introduce an innovative loss function for TFD-Net, named Transformer Deviation Loss Function (TFD-Loss), which can adequately incorporate prior knowledge of samples into the network training process, addressing the issue of imbalanced samples, and thereby enhancing the detection efficiency. Experimental results on public benchmark datasets demonstrate that TFD-Net substantially outperforms other DL-based methods in weakly supervised anomaly detection task.