@inproceedings{7061a8c334c1479d9b2b8dae562683d5,
title = "Source model selection for transfer learning of image classification using supervised contrastive loss",
abstract = "Transfer learning is a framework that improves performance of target task by transferring knowledge from training source task. As deep learning research accumulate, more source models can be easily obtained. In time series domain, the Mean Silhouette Coefficient of the set of feature vectors which forward propagated through source models is used to select the best source model performs target task. But for image classification, the model which is better at generalization could have the lower coefficient. To adjust this, we propose to use another measure, Supervised Contrastive Loss. In this work, we evaluate which measure is better to select the best model. We present the superiority of using the supervised contrastive loss through the comparative experiment.",
keywords = "Contrastive Learning, Deep Learning, Image Classification, Model Selection, Transfer Learning",
author = "Cho, \{Young Seong\} and Samuel Kim and Lee, \{Jee Hyong\}",
note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; 2021 IEEE International Conference on Big Data and Smart Computing, BigComp 2021 ; Conference date: 17-01-2021 Through 20-01-2021",
year = "2021",
month = jan,
doi = "10.1109/BigComp51126.2021.00070",
language = "English",
series = "Proceedings - 2021 IEEE International Conference on Big Data and Smart Computing, BigComp 2021",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "325--329",
editor = "Herwig Unger and Jinho Kim and U Kang and Chakchai So-In and Junping Du and Walid Saad and Young-guk Ha and Christian Wagner and Julien Bourgeois and Chanboon Sathitwiriyawong and Hyuk-Yoon Kwon and Carson Leung",
booktitle = "Proceedings - 2021 IEEE International Conference on Big Data and Smart Computing, BigComp 2021",
}