Publications

Learn Discrimination From Contaminated Data: Multi-Instance Learning With Convex Representation for Unsupervised Anomaly Detection

• Author: Hu W., Liu Y.,Xu J., Chen X., Wang G.
• Abstract:
  • Propsed anomaly discrimination learning with pseudo normality score generation and multi-instancecontrastive learning for distinguishing abnormal samples from the unlabeled set.
  • Combined the discrimination learning with mean-shifted contrastive learning and distribution-shiftingtransformation classification into a multitask representation learning framework to improve the stability androbustness of the training.
• status: Submitted to ICME 2022.