Domain adaptive fusion for adaptive image classification

Andrew Dudley, Bhadrinath Nagabandi, Hemanth Venkateswara, Sethuraman Panchanathan

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Scopus citations


Recent works in the development of deep adaptation networks have yielded progressive improvement on unsupervised domain adaptive classification tasks by reducing the distribution discrepancy between source and target domains. In parallel, the unification of dominant semi-supervised learning techniques has illustrated the unprecedented potential for utilizing unlabeled data to train a classification model in defiance of a discouragingly meager labeled dataset. In this paper, we propose Domain Adaptive Fusion (DAF), a novel domain adaptation algorithm that encourages a domain-invariant linear relationship between the pixel-space of different domains and the prediction-space while being trained under a domain adversarial signal. The thoughtful combination of key components in unsupervised domain adaptation and semi-supervised learning enable DAF to effectively bridge the gap between source and target domains. Experiments performed on computer vision benchmark datasets for domain adaptation endorse the efficacy of our hybrid approach, outperforming all of the baseline architectures on most of the transfer tasks.

Original languageEnglish (US)
Title of host publicationSmart Multimedia - 2nd International Conference, ICSM 2019, Revised Selected Papers
EditorsTroy McDaniel, Stefano Berretti, Igor D.D. Curcio, Anup Basu
Number of pages15
ISBN (Print)9783030544065
StatePublished - 2020
Event2nd International Conference on Smart Multimedia, ICSM 2019 - San Diego, United States
Duration: Dec 16 2019Dec 18 2019

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12015 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349


Conference2nd International Conference on Smart Multimedia, ICSM 2019
Country/TerritoryUnited States
CitySan Diego


  • Domain adaptation
  • Domain-shift
  • Entropy regularization
  • Semi supervised learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • General Computer Science


Dive into the research topics of 'Domain adaptive fusion for adaptive image classification'. Together they form a unique fingerprint.

Cite this