Logo image
Towards a Universal Appearance for Domain Generalization via Adversarial Learning
Conference paper   Peer reviewed

Towards a Universal Appearance for Domain Generalization via Adversarial Learning

Yujui Chen, Tse-Wei Lin and Chiou-Ting Hsu
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Vol.12046 LNCS, pp.297-311
2020

Abstract

Adversarial learning Appearance generalizer Domain generalization Gaussian mixture models Image classification Theoretical Computer Science Computer Science (all)
Domain generalization aims to learn a generalized feature space across multiple source domains so as to adapt the representation to an unseen target domain. In this paper, we focus on two main issues of domain generalization for image classification. First, instead of focusing on aligning data distributions among source domains, we aim to leverage the generalization capability by explicitly referring to image content and style from different source domains. We propose a novel appearance generalizer to learn a universal appearance, which captures domain-invariant content but exhibits various styles. Second, to tackle the class-discriminative issue, we resort to the prominent adversary learning and impose an additional constraint on the discriminator to boost the class-discriminative capability. Experimental results on several benchmark datasets show that the proposed method outperforms existing methods significantly.

Metrics

1 Record Views

Details

Logo image