Invariant Representations without Adversarial Training
This addresses the problem of removing biases and controlling covariates in machine learning models, offering a more efficient alternative to adversarial approaches.
The paper tackles the challenge of learning invariant representations without adversarial training by proposing a single information-theoretic objective, demonstrating that it matches or exceeds state-of-the-art adversarial methods in tasks like fair representation learning and generative modeling.
Representations of data that are invariant to changes in specified factors are useful for a wide range of problems: removing potential biases in prediction problems, controlling the effects of covariates, 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; we instead cast invariant representation learning 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.