Gradient-Guided Annealing for Domain Generalization

TL;DR

This paper introduces Gradient-Guided Annealing (GGA), a novel training method that aligns gradients across domains during early training to improve model robustness and generalization to unseen data distributions.

Contribution

The paper proposes GGA, a new algorithm that enhances domain generalization by iteratively annealing model parameters to align gradients across domains, leading to better robustness.

Findings

GGA achieves state-of-the-art results on five benchmarks.

GGA improves existing domain generalization algorithms when combined.

GGA enhances robustness against domain shifts.

Abstract

Domain Generalization (DG) research has gained considerable traction as of late, since the ability to generalize to unseen data distributions is a requirement that eludes even state-of-the-art training algorithms. In this paper we observe that the initial iterations of model training play a key role in domain generalization effectiveness, since the loss landscape may be significantly different across the training and test distributions, contrary to the case of i.i.d. data. Conflicts between gradients of the loss components of each domain lead the optimization procedure to undesirable local minima that do not capture the domain-invariant features of the target classes. We propose alleviating domain conflicts in model optimization, by iteratively annealing the parameters of a model in the early stages of training and searching for points where gradients align between domains. By discovering a set of parameter values where gradients are updated towards the same direction for each data distribution present in the training set, the proposed Gradient-Guided Annealing (GGA) algorithm encourages models to seek out minima that exhibit improved robustness against domain shifts. The efficacy of GGA is evaluated on five widely accepted and challenging image classification domain generalization benchmarks, where its use alone is able to establish highly competitive or even state-of-the-art performance. Moreover, when combined with previously proposed domain-generalization algorithms it is able to consistently improve their effectiveness by significant margins.