SAND-mask: An Enhanced Gradient Masking Strategy for the Discovery of Invariances in Domain Generalization

TL;DR

This paper introduces SAND-mask, a gradient masking strategy that enhances the discovery of invariant features across domains, improving domain generalization performance especially on Colored MNIST.

Contribution

The paper proposes a novel gradient agreement-based masking technique called SAND-mask to better identify invariances across domains for improved generalization.

Findings

Achieves state-of-the-art accuracy on Colored MNIST.

Provides competitive results on other domain generalization benchmarks.

Enhances invariant feature discovery through gradient agreement masking.

Abstract

A major bottleneck in the real-world applications of machine learning models is their failure in generalizing to unseen domains whose data distribution is not i.i.d to the training domains. This failure often stems from learning non-generalizable features in the training domains that are spuriously correlated with the label of data. To address this shortcoming, there has been a growing surge of interest in learning good explanations that are hard to vary, which is studied under the notion of Out-of-Distribution (OOD) Generalization. The search for good explanations that are \textit{invariant} across different domains can be seen as finding local (global) minimas in the loss landscape that hold true across all of the training domains. In this paper, we propose a masking strategy, which determines a continuous weight based on the agreement of gradients that flow in each edge of network, in order to control the amount of update received by the edge in each step of optimization. Particularly, our proposed technique referred to as "Smoothed-AND (SAND)-masking", not only validates the agreement in the direction of gradients but also promotes the agreement among their magnitudes to further ensure the discovery of invariances across training domains. SAND-mask is validated over the Domainbed benchmark for domain generalization and significantly improves the state-of-the-art accuracy on the Colored MNIST dataset while providing competitive results on other domain generalization datasets.