Domain Adaptation (DA) aims to close the gap in classification performance between the source domain used for training and the target domain where testing takes place. Such performance loss is often encountered due to the dataset bias attributed to changes in the data distributions between the source and target. This seminar begins with a taxonomy of domain adaptation methods, e.g. closed set, open set, unsupervised or semi-supervised. We then present an overview of DA techniques utilizing manifold learning and deep learning, including our methods on manifold-aligned domain adaptation and targeted adversarial domain adaptation. Finally, we address a new DA scenario where the target domain samples are not available all at once, but are acquired in small batches over time and adaptation takes place continually. Our Continual Domain Adaptation methods unitilize concepts from both DA and continual learning. Experiments on various datasets demonstrate the benefits of Continual Domain Adaptation under challenging, data-constrained conditions.
Andreas Savakis is Professor of Computer Engineering at Rochester Institute of Technology (RIT) and Director of the Center for Human-aware Artificial Intelligence (CHAI). He received the B.S. and M.S. degrees in Electrical Engineering from Old Dominion University in Virginia, and the Ph.D. in Electrical and Computer Engineering with Mathematics Minor from North Carolina State University. He was Senior Research Scientist with the Kodak Research Labs before joining RIT. At RIT, he served as department head of Computer Engineering for 10 years and founded the Vision and Image Processing Lab. His research interests include computer vision, deep learning, domain adaptation, human pose estimation, robust and efficient learning, visual tracking, and scene analysis. Dr. Savakis has co-authored over 120 publications and holds 12 U.S. patents. He became Fellow of the American Council on Education (ACE) and received the NYSTAR Technology Transfer Award for Economic Impact, the IEEE Region 1 Award for Outstanding Teaching and the RIT Trustees Scholarship Award.