Transferring the "What" from Experiences
- Ying Wei, Tencent AI Lab
- Time: 2018-04-02 11:15
- Host: Prof. Baoquan Chen
- Venue: Room 101, Courtyard No.5, Jingyuan
Representation learning plays a crucial role in solving mundane to complex tasks in Artificial Intelligence. Psychologists hypothesized and verified that human beings learn representations from past experiences which are organized as clusters – representations are learned within each cluster and generalized across different clusters. In machine learning, the within-cluster representation learning, e.g., deep learning, has achieved huge success, while it fails on the clusters with insufficient labeled data. Transfer learning, as an effective remedy, discovers and leverages knowledge from the other clusters with abundant data to facilitate representation learning in these clusters. In this talk, I will introduce our efforts towards how to formulate and extract the transferable knowledge. I will also present a novel transfer learning framework which we propose to automatically determine the transferable knowledge by implicitly modelling cluster-to-cluster distances. Some successful applications of transfer learning we have explored are also covered.
Ying Wei is a Senior Researcher at Tencent AI Lab. She works on machine learning, and is especially interested in solving challenges in transfer learning by pushing the boundaries of both theories and applications. She received her Ph.D. degree from Hong Kong University of Science and Technology in 2017, under the supervision of Prof. Qiang Yang. Before this, she completed her Bachelor degree from Huazhong University of Science and Technology in 2012.