Understanding the limitations of self-supervised learning for tabular anomaly detection
원문 초록
While self-supervised learning has improved anomaly detection in computer vision and natural language processing, it is unclear whether tabular data can benefit from it. This paper explores the limitations of self-supervision for tabular anomaly detection. We conduct several experiments spanning various pretext tasks on 26 benchmark datasets to understand why this is the case. Our results confirm representations derived from self-supervision do not improve tabular anomaly detection performance compared to using the raw representations of the data. We show this is due to neural networks introducing irrelevant features, which reduces the effectiveness of anomaly detectors. However, we demonstrate that using a subspace of the neural network's representation can recover performance.
함께 살펴볼 논문
공통 태그를 근거로 연결했습니다. 실제 인용 관계나 논문의 우수성을 뜻하지 않습니다.
Self-supervised Learning of Echocardiographic Video Representations via Online Cluster Distillation
A Semi-Supervised Learning Method for the Identification of Bad Exposures in Large Imaging Surveys
Multi-Scale Patch-Based Representation Learning for Image Anomaly Detection and Segmentation
Self-supervised Sparse Representation for Video Anomaly Detection
Weakly and Partially Supervised Learning Frameworks for Anomaly Detection
Chi-Square Wavelet Graph Neural Networks for Heterogeneous Graph Anomaly Detection
토론
이 논문과 연결된 포럼 글이 아직 없습니다.