Exploiting Local Feature Patterns for Unsupervised Domain Adaptation

  • Jun Wen Zhejiang University
  • Risheng Liu Dalian University of Technology
  • Nenggan Zheng Zhejiang University
  • Qian Zheng Zhejiang University
  • Zhefeng Gong Zhejiang University
  • Junsong Yuan State University of New York at Buffalo


Unsupervised domain adaptation methods aim to alleviate performance degradation caused by domain-shift by learning domain-invariant representations. Existing deep domain adaptation methods focus on holistic feature alignment by matching source and target holistic feature distributions, without considering local features and their multi-mode statistics. We show that the learned local feature patterns are more generic and transferable and a further local feature distribution matching enables fine-grained feature alignment. In this paper, we present a method for learning domain-invariant local feature patterns and jointly aligning holistic and local feature statistics. Comparisons to the state-of-the-art unsupervised domain adaptation methods on two popular benchmark datasets demonstrate the superiority of our approach and its effectiveness on alleviating negative transfer.

AAAI Technical Track: Machine Learning