TY - GEN
T1 - Improving maximum classifier discrepancy by considering joint distribution for domain adaptation
AU - Lin, Zehang
AU - Yang, Zhenguo
AU - Situ, Runwei
AU - Huang, Feitao
AU - Lv, Jianming
AU - Li, Qing
AU - Liu, Wenyin
PY - 2018/1/1
Y1 - 2018/1/1
N2 - 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.
AB - 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.
KW - Domain adaptation
KW - Generative Adversarial Networks
KW - Multimedia
KW - Social media
KW - Transfer learning
UR - http://www.scopus.com/inward/record.url?scp=85055934734&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-02925-8_18
DO - 10.1007/978-3-030-02925-8_18
M3 - Conference article published in proceeding or book
AN - SCOPUS:85055934734
SN - 9783030029241
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 253
EP - 268
BT - Web Information Systems Engineering – WISE 2018 - 19th International Conference, 2018, Proceedings
A2 - Wang, Hua
A2 - Zhou, Rui
A2 - Paik, Hye-Young
A2 - Hacid, Hakim
A2 - Cellary, Wojciech
PB - Springer-Verlag
T2 - 19th International Conference on Web Information Systems Engineering, WISE 2018
Y2 - 12 November 2018 through 15 November 2018
ER -