Supervised Domain Adaptation: A Graph Embedding Perspective and a Rectified Experimental Protocol

Lukas Hedegaard Morsing, Omar Ali Sheikh-Omar, Alexandros Iosifidis

Research output: Contribution to journal/Conference contribution in journal/Contribution to newspaperJournal articleResearchpeer-review

21 Citations (Scopus)

Abstract

Domain Adaptation is the process of alleviating distribution gaps between data from different domains. In this paper, we show that Domain Adaptation methods using pair-wise relationships between source and target domain data can be formulated as a Graph Embedding in which the domain labels are incorporated into the structure of the intrinsic and penalty graphs. Specifically, we analyse the loss functions of three existing state-of-the-art Supervised Domain Adaptation methods and demonstrate that they perform Graph Embedding. Moreover, we highlight some generalisation and reproducibility issues related to the experimental setup commonly used to demonstrate the few-shot learning capabilities of these methods. To assess and compare Supervised Domain Adaptation methods accurately, we propose a rectified evaluation protocol, and report updated benchmarks on the standard datasets Office31 (Amazon, DSLR, and Webcam), Digits (MNIST, USPS, SVHN, and MNIST-M) and VisDA (Synthetic, Real).
Original languageEnglish
JournalIEEE Transactions on Image Processing
Volume30
Pages (from-to)8619-8631
Number of pages13
ISSN1057-7149
DOIs
Publication statusPublished - Oct 2021

Keywords

  • domain adaptation
  • few-shot learning
  • graph embedding
  • supervised domain adaptation
  • transfer learning

Fingerprint

Dive into the research topics of 'Supervised Domain Adaptation: A Graph Embedding Perspective and a Rectified Experimental Protocol'. Together they form a unique fingerprint.

Cite this