【15-26期VALSE Webinar活动】 报告嘉宾：付彦伟（Disney research pittsburgh）
主持人：程明明（南开大学）
报告题目：Transductive Multi-View Zero-Shot Learning [Slides]
报告时间：2015年8月19日晚21:00（北京时间）
文章信息：
[1] Transductive Multi-view Embedding for Zero-Shot Recognition and Annotation, ECCV 2014, Yanwei Fu, Timothy M. Hospedales, Tao Xiang, Zhenyong Fu and Shaogang Gong;
[2] Transductive multi-view zero-shot learning, IEEE TPAMI to appear, Yanwei Fu, Timothy M. Hospedales, Tao Xiang, and Shaogang Gong
报告摘要： Attribute learning is emerged as a promising paradigm for bridging the semantic gap and addressing data sparsity via transferring attribute knowledge in image object recognition and relatively simple human action classification. However, attributes are very limited in understanding more complex image/video classes. In this talk, I will introduce our approaches for generalising the previous attribute learning framework to transductive multi-view semantic embedding for zero-shot learning. Specifically, we identify and solve three challenging problems in canonical zero-shot learning pipeline, i.e. projection domain shift problem, prototype sparsity problem and multiple semantic representations embedding problem. Our framework greatly improves the zero-shot learning accuracy on several benchmark dataset.
报告人简介： Yanwei Fu received the PhD degree from Queen Mary University of London in 2014, and the MEng degree from the Department of Computer Science & Technology, Nanjing University in 2011, China. He is a postdoctoral researcher with Leonid Sigal in Disney Research, Pittsburgh, which is co-located with Carnegie Mellon University. His research interests include image and video understanding and description, robust ranking and learning to rank, and large-scale surveillance video analysis. | 