Constraint-Guided Autoencoders to Enforce a Predefined Threshold on Anomaly Scores:An Application in Machine Condition Monitoring

作     者：Maarten Meire Quinten Van Baelen Ted Ooijevaar Peter Karsmakers 

作者机构：Department of Computer ScienceKU LeuvenADVISE-DTAIKleinhoefstraat 4 B-2440 GeelBelgium Leuven.AI–KU Leuven institute for AI3000 LeuvenBelgium Flanders Make@KU Leuven3000 LeuvenBelgium Flanders Make vzwCoreLab MotionSLeuven 3001Belgium 

出 版 物：《Journal of Dynamics, Monitoring and Diagnostics》 (动力学、监测与诊断学报（英文）)

年 卷 期：2023年第2卷第2期

页      面：144-154页

学科分类：08[工学] 0812[工学-计算机科学与技术（可授工学、理学学位）] 

基　　金：This research received funding from the Flemish Government(AI Research Program) This research has received support of Flanders Make,the strategic research center for the manufacturing industry 

主　　题：anomaly detection autoencoders deep learning 

摘      要：Anomaly detection(AD)is an important task in a broad range of domains.A popular choice for AD are Deep Support Vector Data Description *** learning such models,normal data is mapped close to and anomalous data is mapped far from a center,in some latent space,enabling the construction of a sphere to separate both types of ***,it was observed:(i)that the center and radius of such sphere largely depend on the training data and model initialization which leads to difficulties when selecting a threshold,and(ii)that the center and radius of this sphere strongly impact the model AD performance on unseen *** this work,a more robust AD solution is proposed that(i)defines a sphere with a fixed radius and margin in some latent space and(ii)enforces the encoder,which maps the input to a latent space,to encode the normal data in a small sphere and the anomalous data outside a larger sphere,with the same *** results indicate that the proposed algorithm attains higher performance compared to alternatives,and that the difference in size of the two spheres has a minor impact on the performance.