Zero Shot Domain Generalization

TL;DR

This paper introduces Zero-Shot Domain Generalization, a new setting where models must generalize to unseen domains and classes using semantic information, extending traditional domain generalization to more challenging scenarios.

Contribution

The paper proposes a simple strategy leveraging semantic class information to adapt existing domain generalization methods for zero-shot scenarios, and establishes a baseline on multiple datasets.

Findings

Effective adaptation of DG methods to zero-shot settings.

Strong baseline results on CIFAR-10, CIFAR-100, F-MNIST, and PACS.

First effort to address zero-shot domain and class generalization.

Abstract

Standard supervised learning setting assumes that training data and test data come from the same distribution (domain). Domain generalization (DG) methods try to learn a model that when trained on data from multiple domains, would generalize to a new unseen domain. We extend DG to an even more challenging setting, where the label space of the unseen domain could also change. We introduce this problem as Zero-Shot Domain Generalization (to the best of our knowledge, the first such effort), where the model generalizes across new domains and also across new classes in those domains. We propose a simple strategy which effectively exploits semantic information of classes, to adapt existing DG methods to meet the demands of Zero-Shot Domain Generalization. We evaluate the proposed methods on CIFAR-10, CIFAR-100, F-MNIST and PACS datasets, establishing a strong baseline to foster interest in this new research direction.