Unsupervised Model Selection for Variational Disentangled Representation Learning

TL;DR

This paper introduces UDR, an unsupervised method for selecting disentangled models without ground truth labels, enabling better hyperparameter tuning and model comparison in complex domains.

Contribution

It proposes a novel unsupervised ranking method for disentangled representations based on recent theoretical insights, outperforming supervised methods in model selection tasks.

Findings

UDR performs comparably to supervised methods on 5,400 models.

The ranking correlates well with task performance across domains.

The approach is robust and applicable to various model classes.

Abstract

Disentangled representations have recently been shown to improve fairness, data efficiency and generalisation in simple supervised and reinforcement learning tasks. To extend the benefits of disentangled representations to more complex domains and practical applications, it is important to enable hyperparameter tuning and model selection of existing unsupervised approaches without requiring access to ground truth attribute labels, which are not available for most datasets. This paper addresses this problem by introducing a simple yet robust and reliable method for unsupervised disentangled model selection. Our approach, Unsupervised Disentanglement Ranking (UDR), leverages the recent theoretical results that explain why variational autoencoders disentangle (Rolinek et al, 2019), to quantify the quality of disentanglement by performing pairwise comparisons between trained model representations. We show that our approach performs comparably to the existing supervised alternatives across 5,400 models from six state of the art unsupervised disentangled representation learning model classes. Furthermore, we show that the ranking produced by our approach correlates well with the final task performance on two different domains.