[AIP Distinguished Lecture] Prof. Hsuan-Tien Lin (National Taiwan University) "Connecting Domain Adaptation to Other Learning Problems"

A RIKEN AIP Distinguished Lecture by Prof. Hsuan-Tien Lin of National Taiwan University, held in hybrid form online and at the open space of the RIKEN Nihonbashi Office. The talk argues that domain adaptation (DA) is often a known machine learning problem in disguise. He presents two works from his group: Source Label Adaptation (CVPR 2023), which treats source labels in semi-supervised DA as noisy versions of ideal target labels, and a study of the "extreme" universal DA regime (ICML 2025), where alignment and uniformity objectives from self-supervised learning fix dimensional collapse of target representations. Prof. Lin is a professor at NTU and a former Chief Data Scientist at Appier.