Date: Monday 9th February 2014 at 10:00AM
Abstract: We will present learning approaches to predicting end-to-end network performance such as round-trip time and available bandwidth. The key is to formulate the prediction problem as a problem of matrix completion. A novel concept is the rating of network paths, i.e. acquiring quantized measures of path properties, which is appealing in that ratings are not only informative but also cheap to obtain.
Short Bio:
Yongjun is currently a postdoc at Department of Electrical Engineering and Computer Science, University of Liège (ULG), Belgium. She received her B.S. in 1999 and M.S. in 2002, both from Department of Computer Science, Guangxi University, China. She joined Research Unit of Networking (RUN) at ULG, led by Professor Guy Leduc, in 2007, and defended her PhD thesis in January 2013. Before, she had worked as a software engineer in a small computer company in Beijing, China.
