Are Disentangled Representations Helpful for Abstract Visual Reasoning?
This addresses the problem of evaluating the practical utility of disentangled representations for AI researchers, providing empirical evidence that supports their claimed benefits.
The paper investigates whether disentangled representations improve performance on abstract visual reasoning tasks, finding that they lead to better downstream performance, specifically enabling quicker learning with fewer samples.
A disentangled representation encodes information about the salient factors of variation in the data independently. Although it is often argued that this representational format is useful in learning to solve many real-world down-stream tasks, there is little empirical evidence that supports this claim. In this paper, we conduct a large-scale study that investigates whether disentangled representations are more suitable for abstract reasoning tasks. Using two new tasks similar to Raven's Progressive Matrices, we evaluate the usefulness of the representations learned by 360 state-of-the-art unsupervised disentanglement models. Based on these representations, we train 3600 abstract reasoning models and observe that disentangled representations do in fact lead to better down-stream performance. In particular, they enable quicker learning using fewer samples.