Learning Causal Semantic Representation for Out-of-Distribution Prediction

TL;DR

This paper introduces a causal generative model that separates semantic and variation factors, enabling better out-of-distribution prediction and generalization from a single training domain.

Contribution

The paper proposes a novel causal semantic generative model with a variational Bayes approach, improving OOD prediction and providing theoretical guarantees for semantic factor identification.

Findings

Enhanced OOD prediction performance over baselines

Theoretical proof of semantic factor identification under certain conditions

Bounded OOD generalization error and successful adaptation

Abstract

Conventional supervised learning methods, especially deep ones, are found to be sensitive to out-of-distribution (OOD) examples, largely because the learned representation mixes the semantic factor with the variation factor due to their domain-specific correlation, while only the semantic factor causes the output. To address the problem, we propose a Causal Semantic Generative model (CSG) based on a causal reasoning so that the two factors are modeled separately, and develop methods for OOD prediction from a single training domain, which is common and challenging. The methods are based on the causal invariance principle, with a novel design in variational Bayes for both efficient learning and easy prediction. Theoretically, we prove that under certain conditions, CSG can identify the semantic factor by fitting training data, and this semantic-identification guarantees the boundedness of OOD generalization error and the success of adaptation. Empirical study shows improved OOD performance over prevailing baselines.