Improving Generalization for Abstract Reasoning Tasks Using Disentangled Feature Representations

TL;DR

This paper demonstrates that disentangled VAE-based unsupervised learning significantly improves generalization in relational reasoning tasks derived from Raven Progressive Matrices compared to supervised training.

Contribution

It introduces the use of disentangled VAE's for unsupervised learning to enhance generalization in abstract reasoning tasks, outperforming supervised methods.

Findings

Unsupervised disentangled VAE's outperform supervised models in generalization.

Proper objective functions are crucial for effective latent space learning.

Disentangled representations improve reasoning across relational problems.

Abstract

In this work we explore the generalization characteristics of unsupervised representation learning by leveraging disentangled VAE's to learn a useful latent space on a set of relational reasoning problems derived from Raven Progressive Matrices. We show that the latent representations, learned by unsupervised training using the right objective function, significantly outperform the same architectures trained with purely supervised learning, especially when it comes to generalization.