期刊
IEEE TRANSACTIONS ON IMAGE PROCESSING
卷 30, 期 -, 页码 8008-8018出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TIP.2021.3112012
关键词
Adaptation models; Training; Collaboration; Feature extraction; Computational modeling; Head; Neural networks; Domain adaptation; domain generalization; collaborative ensemble learning
资金
- National Natural Science Foundation of China [61876176, U1713208]
- National Key Research and Development Program of China [2020YFC2004800]
- Science and Technology Service Network Initiative of Chinese Academy of Sciences [KFJ-STS-QYZX-092]
- Shanghai Committee of Science and Technology, China [20DZ1100800]
The study focuses on generalizing deep neural networks from multiple source domains to a target domain, and proposes a unified framework called DAEL, which aims to improve accuracy on unseen target domains by collaboratively learning experts.
The problem of generalizing deep neural networks from multiple source domains to a target one is studied under two settings: When unlabeled target data is available, it is a multi-source unsupervised domain adaptation (UDA) problem, otherwise a domain generalization (DG) problem. We propose a unified framework termed domain adaptive ensemble learning (DAEL) to address both problems. A DAEL model is composed of a CNN feature extractor shared across domains and multiple classifier heads each trained to specialize in a particular source domain. Each such classifier is an expert to its own domain but a non-expert to others. DAEL aims to learn these experts collaboratively so that when forming an ensemble, they can leverage complementary information from each other to be more effective for an unseen target domain. To this end, each source domain is used in turn as a pseudo-target-domain with its own expert providing supervisory signal to the ensemble of non-experts learned from the other sources. To deal with unlabeled target data under the UDA setting where real expert does not exist, DAEL uses pseudo labels to supervise the ensemble learning. Extensive experiments on three multi-source UDA datasets and two DG datasets show that DAEL improves the state of the art on both problems, often by significant margins.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据