Best sources forward: domain generalization through source-specific nets

TL;DR

This paper introduces a novel domain generalization method using a deep network with source-specific classifiers and a domain-agnostic component, effectively handling multiple source domains to improve test-time generalization.

Contribution

It proposes a new approach with multiple domain-specific classifiers and a domain-agnostic component for better domain generalization across multiple sources.

Findings

Outperforms existing DG methods on benchmark datasets.

Effectively estimates source probabilities for target samples.

Improves generalization to unseen domains.

Abstract

A long standing problem in visual object categorization is the ability of algorithms to generalize across different testing conditions. The problem has been formalized as a covariate shift among the probability distributions generating the training data (source) and the test data (target) and several domain adaptation methods have been proposed to address this issue. While these approaches have considered the single source-single target scenario, it is plausible to have multiple sources and require adaptation to any possible target domain. This last scenario, named Domain Generalization (DG), is the focus of our work. Differently from previous DG methods which learn domain invariant representations from source data, we design a deep network with multiple domain-specific classifiers, each associated to a source domain. At test time we estimate the probabilities that a target sample belongs to each source domain and exploit them to optimally fuse the classifiers predictions. To further improve the generalization ability of our model, we also introduced a domain agnostic component supporting the final classifier. Experiments on two public benchmarks demonstrate the power of our approach.