Learning Independent Features with Adversarial Nets for Non-linear ICA

TL;DR

This paper introduces an adversarial learning approach to extract independent features for ICA, bypassing the need for explicit dependence measures like mutual information, and demonstrates its effectiveness on linear and non-linear ICA tasks.

Contribution

It proposes a novel adversarial framework for learning independent features that implicitly optimize dependence measures without density estimation, applicable to various model architectures.

Findings

Effective in solving linear ICA problems

Applicable to non-linear ICA tasks

Flexible across different model architectures

Abstract

Reliable measures of statistical dependence could be useful tools for learning independent features and performing tasks like source separation using Independent Component Analysis (ICA). Unfortunately, many of such measures, like the mutual information, are hard to estimate and optimize directly. We propose to learn independent features with adversarial objectives which optimize such measures implicitly. These objectives compare samples from the joint distribution and the product of the marginals without the need to compute any probability densities. We also propose two methods for obtaining samples from the product of the marginals using either a simple resampling trick or a separate parametric distribution. Our experiments show that this strategy can easily be applied to different types of model architectures and solve both linear and non-linear ICA problems.