Institute of Information Science, Academia Sinica

 Deep neural networks have demonstrated state-of-the-art performance on a wide range of tasks. These models, however, often have difficulty in generalizing to unseen samples and domains. The major challenges include representation biases in our datasets, scarcity of labeled training examples in target domains, and the diversity of testing cases. In this talk, I will show how we tackle these challenges and learn models that generalize better to unseen testing cases and domains. I will discuss several techniques for improving generalization in the context of specific computer vision applications, including activity understanding, object recognition, and semantic segmentation.

 Jia-Bin Huang is an Assistant Professor in the Bradley Department of Electrical and Computer Engineering at Virginia Tech. He received a B.S. degree in Electronics Engineering from National Chiao-Tung University, Hsinchu, Taiwan, and his Ph.D. degree in the Department of Electrical and Computer Engineering at the University of Illinois, Urbana-Champaign. His research interests include computer vision, computer graphics, and machine learning. He is the recipient of a Google Faculty Award, SAMSUNG Global Outreach Award, and an NSF CRII Award.