Learning Conditional Invariance through Cycle Consistency

TL;DR

This paper introduces a novel cycle consistency approach with separate latent subspaces to learn invariant and sparse representations, improving interpretability and invariance in deep latent models.

Contribution

It proposes a new method using cycle consistency and semantic knowledge to enforce invariance and sparsity in latent spaces, allowing continuous properties and better model selection.

Findings

Identifies more meaningful factors in synthetic and molecular data.

Produces sparser and more interpretable models.

Enhances invariance properties in learned representations.

Abstract

Identifying meaningful and independent factors of variation in a dataset is a challenging learning task frequently addressed by means of deep latent variable models. This task can be viewed as learning symmetry transformations preserving the value of a chosen property along latent dimensions. However, existing approaches exhibit severe drawbacks in enforcing the invariance property in the latent space. We address these shortcomings with a novel approach to cycle consistency. Our method involves two separate latent subspaces for the target property and the remaining input information, respectively. In order to enforce invariance as well as sparsity in the latent space, we incorporate semantic knowledge by using cycle consistency constraints relying on property side information. The proposed method is based on the deep information bottleneck and, in contrast to other approaches, allows using continuous target properties and provides inherent model selection capabilities. We demonstrate on synthetic and molecular data that our approach identifies more meaningful factors which lead to sparser and more interpretable models with improved invariance properties.