Deep Feature Selection for Anomaly Detection Based on Pretrained Network and Gaussian Discriminative Analysis

Abstract

Deep learning neural network serves as a powerful tool for visual anomaly detection and fault diagnosis, attributed to its strong abstractive interpretation ability in the representation domain. The deep features from neural networks that are pre-trained on the ImageNet classification task have been proved to be useful for anomaly detection based on Gaussian discriminant analysis. However, with the ever-increasing complexity of deep learning neural networks, the set of deep features becomes massive where redundancy appears to be inevitable. The redundant features increase computational cost and degrade the performance of the anomaly detection method. In this paper, we discuss the deep feature selection for the anomaly detection task and show how to reduce the redundancy in the representation domain. We propose a horizontal selection (dimensional reduction) method of features with subspace decomposition and a vertical selection to identify the most effective network layer for anomaly detection and fault diagnosis. We test the proposed method on two public datasets, one for anomaly detection task and one for fault diagnosis of bearings. We show the significance of different network layers and feature subspaces on anomaly detection tasks, and prove the effectiveness of the feature selection strategy.