Abstract
Autoencoders (AEs) have been widely used for unsupervised anomaly detection. They learn from normal samples such that they produce high reconstruction errors for anomalous samples. However, AEs can exhibit the over-detection issue because they imperfectly reconstruct not only anomalous samples but also normal ones. To address this issue, we introduce an outlier-exposed style distillation network (OE-SDN) that mimics the mild distortions caused by an AE, which are termed as style translation. We use the difference between the outputs of the OE-SDN and AE as an alternative anomaly score. Experiments on anomaly classification and segmentation tasks show that the performance of our method is superior to existing methods.
Highlights
The objective of unsupervised anomaly detection is to identify anomalous samples from data
UNSUPERVISED ANOMALY CLASSIFICATION We evaluated our method for the MNIST [22] and CIFAR10 [23] datasets for the classification task of unsupervised anomaly detection
We investigated the effectiveness of the proposed method in comparison with three unsupervised anomaly segmentation methods that were used as baselines in [4]: AnoGAN [8], a method based on convolutional neural network (CNN) feature similarity [28], and an AE [4] with an alternative architecture
Summary
The objective of unsupervised anomaly detection is to identify anomalous samples from data. Unsupervised anomaly detection assumes that only normal samples are present while anomalous samples are absent in the training dataset. These extreme distortions are regarded as content translation. RECONSTRUCTION-BASED UNSUPERVISED ANOMALY DETECTION A prevalent choice for anomaly detection is reconstructionbased anomaly detection using such models as an AE [5], a variational autoencoder (VAE) [6], and a generative adversarial network (GAN) [8] It identifies a sample as an anomaly if the reconstruction error is above a certain threshold. The proposed method calculates the anomaly score by comparing the AE and OE-SDN outputs
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
