SepVAE: A Contrastive VAE to Separate Pathological Patterns from Healthy Ones

Abstract

Contrastive Analysis VAE (CA-VAEs) is a family of Variational auto-encoders (VAEs) that aims at separating the common factors of variation between a \textit{background} dataset (BG) (\textit{i.e.,} healthy subjects) and a \textit{target} dataset (TG) (\textit{i.e.,} patients) from the ones that only exist in the target dataset. To do so, these methods separate the latent space into a set of \textbf{salient} features (\textit{i.e.,} proper to the target dataset) and a set of \textbf{common} features (\textit{i.e.,} exist in both datasets). Currently, all CA-VAEs models fail to prevent sharing of information between the latent spaces and to capture all salient factors of variation. To this end, we introduce two crucial regularization losses: a disentangling term between common and salient representations and a classification term between background and target samples in the salient space. We show a better performance than previous CA-VAEs methods on three medical applications and a natural images dataset (CelebA).