Learning Conditional Invariance through Cycle Consistency
This work addresses the problem of identifying interpretable latent factors for researchers in machine learning, offering an incremental improvement over existing methods for enforcing invariance.
The paper tackles the challenge of learning independent factors of variation in datasets by enforcing invariance in latent spaces, resulting in more meaningful, sparse, and interpretable models with improved invariance properties on synthetic and molecular data.
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.