Speaker: Wenjun (Kevin) Zeng   http://people.cs.missouri.edu/~zengw/
Time:    2:30 pm, Jul 22, 2013, Monday
Venue:   信息学院一楼会议室

Biography: Wenjun (Kevin) Zeng is a Fellow of the IEEE. He is a Full Professor with the Computer Science Department of University of Missouri, Columbia, MO. He received his B.E., M.S., and Ph.D. degrees from Tsinghua University, the University of Notre Dame, and Princeton University, respectively, all in electrical engineering. His current research interest includes mobile computing, social media analysis, semantic search, distributed source/video coding, 3-D analysis and coding, multimedia networking, and content and network security.

Topic1: Trend Aware Proactive Caching of Online Video
Abstraction 1: In the recent years, popularity of social media and online video sharing services has grown at an unprecedentedly fast pace. A massive amount of information is being generated and uploaded to the Internet every day. Modern Internet faces new challenges with a growing demand on video. Caching content has been an effective method for improving quality of service for online video. Previous research has discovered that correlation between social media and video-sharing portals exists. In this talk, I will present a mainstream media driven trend detection and caching framework that transits the knowledge of detected trends to online video sharing portals, to detect emerging popular videos, and pre-cache them at strategically deployed caching nodes. In particular, we propose to explore a combination of topic modeling and frequent pattern mining to design a cross-platform video popularity prediction scheme. We further propose a trend-aware and reputation-based video-ranking algorithm to select correct caching candidates among a large array of redundant content for proactive caching by the Internet Service Providers (ISP). Experimental results show that the proposed proactive caching framework can significantly outperform conventional caching methods that are based on historical popularity.

Topic2: Social Multimedia Signals: when Social Networking meets Signal Processing
Abstraction 2: In recent years, social computing has generated momentous interest from researchers in multi-disciplinary fields, including computational social science, multimedia, network science and visualization etc. Social Multimedia refers to media content generated from social networks. A Social Multimedia Signal presumes human users as sensors and contains the spatio-temporal activity pattern of users (or user community) with respect to some multimedia content shared within the social network. Some signals provide perspective about single media content, e.g., 'Likes' on Facebook. There also exist social multimedia signals that provide contextual knowledge regarding the entire social ecosystem within which media can be shared. Twitter trending topics is a fair reflection of the hot topics of discussion in the Twitter world. Social Multimedia Signal Processing aims to transform the noise-like phenomena in social media into signals useful for building novel socially-aware multimedia applications such as cross-domain media recommendations and targeted advertising, exploring new social marketing methods and a fresh way to look at the existence of multimedia in online social networks.