Seminar Title: Adaptive Knowledge Transfer
Abstract: Transfer learning and multi-task learning allow machine-learning models to draw on knowledge from related tasks or domains, particularly when labeled data are limited or expensive to obtain. However, challenges such as data heterogeneity, negative transfer, high dimensionality and data privacy remain.
In this seminar, Yang Feng will present a unified statistical perspective on adaptive knowledge transfer. He will discuss methods for identifying and leveraging informative data sources, representation multi-task learning and emerging approaches to unsupervised and federated transfer learning. The presentation will demonstrate how statistical theory can guide the development of scalable, robust AI systems across interdisciplinary applications.
Registration is free and encouraged for the seminar, learn more at https://events.blackthorn.io/en/2Eo1ku6/g/t6Vv9jWB55/adaptive-knowledge-transfer-with-yang-feng-5a2CPm4MGDV/overview
