Improving maximum classifier discrepancy by considering joint distribution for domain adaptation

Zehang Lin, Zhenguo Yang, Runwei Situ, Feitao Huang, Jianming Lv, Qing Li, Wenyin Liu

Research output: Chapter in book / Conference proceedingConference article published in proceeding or bookAcademic researchpeer-review

3 Citations (Scopus)

Abstract

Recently, domain adaptation has gained great popularity, while most researchers are focusing on domains in homogenous modalities, e.g., image domains. In reality, heterogeneous domains are pretty common and more challenging. In this paper, we present MCD-JD—a Maximum Classifier Discrepancy model which considers the joint distribution of the source and target domain data for heterogeneous domain adaption. MCD-JD derives from Generative Adversarial Networks (GAN) consisting of two parts, i.e., minimizing the discrepancy of joint distribution, and maximizing classifier discrepancy. Specifically, the first part uses the Maximum Mean Discrepancy (MMD) regularization to adapt the data distributions between source and target domains. The second part utilizes two different classifiers to maximize their discrepancy of making predictions on the target domain data, which further minimizes the discrepancy of data distributions between source and target domains. We collect a dataset depicting real-world events (e.g., protests, explosions, etc.) from multiple heterogeneous data domains, including news media textual articles, social media (Flickr) images, and YouTube videos. Extensive experiments conducted on the real-world dataset manifest the effectiveness of MCD-JD, which outperforms state-of-the-art benchmark models.

Original languageEnglish
Title of host publicationWeb Information Systems Engineering – WISE 2018 - 19th International Conference, 2018, Proceedings
EditorsHua Wang, Rui Zhou, Hye-Young Paik, Hakim Hacid, Wojciech Cellary
PublisherSpringer-Verlag
Pages253-268
Number of pages16
ISBN (Print)9783030029241
DOIs
Publication statusPublished - 1 Jan 2018
Externally publishedYes
Event19th International Conference on Web Information Systems Engineering, WISE 2018 - Dubai, United Arab Emirates
Duration: 12 Nov 201815 Nov 2018

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume11234 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference19th International Conference on Web Information Systems Engineering, WISE 2018
Country/TerritoryUnited Arab Emirates
CityDubai
Period12/11/1815/11/18

Keywords

  • Domain adaptation
  • Generative Adversarial Networks
  • Multimedia
  • Social media
  • Transfer learning

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this