- Day 3 (Nov.19), 9:50-10:40
- 2nd floor "WINC Hall"
Training deep neural networks from limited supervised data for constructing an accurate prediction model is one of the crucial tasks in visual recognition problems. In this talk, we introduce domain adaptation methods for both classification and generative models that transfer knowledge in a label rich domain to a label scarce domain. We also present a new learning method using between-class examples to train DNNs and boost a classification performance from limited data. Besides, we will briefly introduce various topics that we are working on in our team.
Tatsuya Harada is a Professor in the Research Center for Advanced Science and Technology at the University of Tokyo. His research interests center on visual recognition, machine learning, and intelligent robot. He received his Ph.D. from the University of Tokyo in 2001. He is also a team leader at RIKEN AIP and a vice director of Research Center for Medical Bigdata at National Institute of Informatics, Japan.