作者: Lorne Swersky , Henrique O Marques , Jöerg Sander , Ricardo JGB Campello , Arthur Zimek
DOI: 10.1109/DSAA.2016.8
关键词:
摘要: It has been shown that unsupervised outlier detection methods can be adapted to the one-class classification problem. In this paper, we focus on comparison of oneclass algorithms with such methods, improving previous studies in several important aspects. We study a number and rigorous experimental setup, comparing them large datasets different characteristics, using performance measures. Our experiments led conclusions do not fully agree those work.