Dr. Qifei Wang, Senior Software Engineer at Google
Berkeley, California, United States
Dr. Qifei Wang received his Ph.D. degrees in Control Engineering and Science in the Department of Automation, Tsinghua University, Beijing, China, in 2013. He worked as an associate researcher in Microsoft Research Asia between 2013 and 2014, and a Post-Doctoral Research Fellow in University of California, Berkeley, between 2014 and 2015. Since 2016, Dr. Qifei Wang joined Google and worked on on-device computer vision and machine learning technologies. His research interests include computer vision, machine learning, multimedia processing and communication.
Speech Topic: Recent advances and applications of on-device multi-domain learning
Deep learning has achieved remarkable success in various areas where training and test data are sampled from the same domain. In real applications, however, there is a great chance of applying a deep learning model to the data from a new domain. Training a dedicated model for each domain is neither data efficient due to the high data acquisition cost nor resource efficient when running the inference on the resource constrained devices. In this talk, we will review the recent progress in the multi-domain learning and present the recent advances in efficient on-device multi-domain learning, including the model architecture and domain generalization, etc. Beyond the technology, this talk will further discuss the applications of multi-domain learning in on-device computer vision systems. Finally, this talk will talk about the future research directions of the on-device multi-domain learning.