2019/12/26 by Pierrick Chatillon, Chatillon, Pierrick, Coloma Ballester +1
Computer Science · Decision Sciences · #Anomaly Detection Techniques and Applications #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #Multidisciplinary Science and Engineering Research #Network Security and Intrusion Detection
paper · pdf · doi:10.48550/arxiv.1912.11843
openalex publication_date 2019/12/26 · openalex created_date 2022/07/19 · openalex updated_date 2026/07/28
Anomaly detection is a difficult problem in many areas and has recently been\nsubject to a lot of attention. Classifying unseen data as anomalous is a\nchallenging matter. Latest proposed methods rely on Generative Adversarial\nNetworks (GANs) to estimate the normal data distribution, and produce an\nanomaly score prediction for any given data. In this article, we propose a\nsimple yet new adversarial method to tackle this problem, denoted as\nHistory-based anomaly detector (HistoryAD). It consists of a self-supervised\nmodel, trained to recognize 'normal' samples by comparing them to samples based\non the training history of a previously trained GAN. Quantitative and\nqualitative results are presented evaluating its performance. We also present a\ncomparison to several state-of-the-art methods for anomaly detection showing\nthat our proposal achieves top-tier results on several datasets.\n