Information-Theoretic Representation Learning for Positive-Unlabeled Classification

Recent advances in weakly supervised classification allow us to train a classifier only from positive and unlabeled (PU) data. However, existing PU methods typically require an accurate estimate of the class-prior probability, which is critical bottleneck particularly for high-dimensional This problem has been commonly addressed by applying principal component analysis advance, but such unsupervised dimension reduction can collapse underlying class structure. In this paper, we propose novel representation learning method data based on information-maximization principle. Our does not estimation thus be used as preprocessing classification. Through experiments, demonstrate that our combined with deep neural networks highly improves accuracy estimation, leading state-of-the-art performance.