Coping with training contamination in unsupervised distributional anomaly detection | IEEE Conference Publication | IEEE Xplore