Towards Shape Biased Unsupervised Representation Learning for Domain Generalization

TL;DR

This paper introduces a novel self-supervised learning framework that enhances shape bias in representations, improving domain generalization without prior domain knowledge, validated across multiple datasets.

Contribution

The proposed framework combines domain diversification and jigsaw puzzles to promote shape bias in self-supervised learning, advancing domain generalization capabilities.

Findings

Outperforms state-of-the-art domain generalization methods

Effective across multiple datasets including PACS, Office-Home, VLCS, and Digits

Enhances shape bias in learned representations

Abstract

It is known that, without awareness of the process, our brain appears to focus on the general shape of objects rather than superficial statistics of context. On the other hand, learning autonomously allows discovering invariant regularities which help generalization. In this work, we propose a learning framework to improve the shape bias property of self-supervised methods. Our method learns semantic and shape biased representations by integrating domain diversification and jigsaw puzzles. The first module enables the model to create a dynamic environment across arbitrary domains and provides a domain exploration vs. exploitation trade-off, while the second module allows the model to explore this environment autonomously. This universal framework does not require prior knowledge of the domain of interest. Extensive experiments are conducted on several domain generalization datasets, namely, PACS, Office-Home, VLCS, and Digits. We show that our framework outperforms state-of-the-art domain generalization methods by a large margin.