Evading the Adversary in Invariant Representation

Representations of data that are invariant to changes in specified nuisance factors are useful for a wide range of problems: removing potential bias in prediction problems, controlling the effects of known confounders, and disentangling meaningful factors of variation. Unfortunately, learning representations that exhibit invariance to arbitrary nuisance factors yet remain useful for other tasks is challenging. Existing approaches cast the trade-off between task performance and invariance in an adversarial way, using an iterative minimax optimization. We show that adversarial training is unnecessary and sometimes counter-productive by casting invariant representation learning for various tasks as a single information-theoretic objective that can be directly optimized. We demonstrate that this approach matches or exceeds performance of state-of-the-art adversarial approaches for learning fair representations and for generative modeling with controllable transformations.