活動介紹：

題目: From “It Doesn’t Work” to “It Works So Well”

主講人: Dr. Liangliang Cao

時間：5月22日上午9:30

地點：軟件所5號樓4層第一會議室 

內容介紹：

In our research lives it often happens that our proposed idea does not work well as we expected. However, such phenomenon often motivates us to design better solutions. In this talk, I would share some experiences of handling non-working ideas. Firstly I will introduce a project on 3D object reconstruction from line drawings. Our initial idea is to change the optimization variable from vertices to planar variables.  However, such a seemly equivalent formulation results in terrible reconstruction errors. By fixing the bug, we design a new algorithm, which was published in IEEE transaction on Pattern Analysis and Machine Intelligence. Following the similar spirit, I will then introduce the Delta-SimRank algorithm, which won the best paper award in the SIGKDD workshop on Big Data Mining. Delta-SimRank was motivated by a failure on Hadoop system, which generates too many intermediate data and brings much traffic in the distributed system. By rewriting SimRank algorithm in another form, we are able to design a new algorithm which significantly reduces the traffics in distributed system and speeds up the computation. These research experiences suggest that we shall not overlook the reason why our idea does not work since we cannot escape from this phenomenon. Sometimes by exploring these bugs we can learn a lot of the unknown world.

主講人介紹：

Dr. Liangliang Cao is a research staff member in IBM T.J. Waston Research Center. He also holds an adjunct faculty position at the Columbia University in the City of New York. His work has received three times the IBM Research Division Award (individual award) as well as one Outstanding Technical Accomplishment (group award). He is now leading the efforts of building big data learning toolkit for IBM Multimedia group. Before joining IBM, he did summer interns at Kodak, Microsoft, and NEC Research Labs and received his Ph.D. from University of Illinois at Urbana-Champaign. He received the best paper award from  the International Workshop on Big Data Mining, 2012 He was listed as an IBM Emerging Leader in Multimedia and Signal Processing in 2010 and Facebook Fellowship Finalist in 2010. He was a general chair of Greater New York Area Multimedia and Vision Meeting from 2012 to 2013. He was also a founding chair of ACM workshop on Geo-Multimedia. He was an area chair of ACM Multimedia 2012 and WACV 2014

活動宣傳：軟件所青聯舉辦計算機視覺與新媒體技術交流研討會