A method for semi supervised representation learning of images.
The intended use of the learned representations is to train a one class classifier (OCC).
[1] Learning and Evaluating Representations for Deep One-Class Classification (arxiv)
[2] Discovering Anomalous Data with Self-Supervised Learning (arxiv)