D2IFLN: Disentangled Domain-Invariant Feature Learning Networks for Domain Generalization

Authors:
Publisher website
Google Scholar
Abstract
Domain generalization (DG) aims to learn a model that generalizes well to an unseen test distribution. Mainstream methods follow the domain-invariant representational learning philosophy to achieve this goal. However, due to the lack of priori knowledge to determine which features are domain-specific and task-independent, and which features are domain-invariant and task-relevant, existing methods typically learn entangled representations, limiting their capacity to generalize to the distribution-shifted target domain. To address this issue, in this paper, we propose novel Disentangled Domain-Invariant Feature Learning Networks (D2IFLN) to realize feature disentanglement and facilitate domain-invariant feature learning. Specifically, we introduce a semantic disentanglement network and a domain disentanglement network, disentangling the learned domaininvariant features from both domain-specific class-irrelevant features and domain-discriminative features. To avoid the semantic confusion in adversarial learning for domain-invariant feature learning, we further introduce a graph neural network to aggregate different domain semantic features during model training. Extensive experiments on three DG benchmarks show that the proposed D2IFLN performs better than the state-of-theart.
Funding Information
  • National Key Research and Development Program of China (2021YFB2501104)
  • Shanghai Municipal Science and Technology Major Project (2018SHZDZX01)
  • ZJ Lab
  • Shanghai Center for Brain Science and Brain-Inspired Technology
  • Shanghai Rising Star Program (21QC1400900)
  • Tongji-Westwell Autonomous Vehicle Joint Lab Project
  • Guided Autonomous Logistics Vehicle Project

This publication has 39 references indexed in Scilit: