主題:    Select representative data to train SVM for big data  classification主講人:   Xiaoou Li 地點:   松江校區一號沐鸣樓140報告廳時間:   2018-11-15 13:00:00組織單位:   計算機沐鸣

主講人簡介：

Prof. Xiaoou Li obtained B. S degree of appliedmathematics in 1991 and PhD  degree of Automatic Control in 1995 fromNortheastern University, Shenyang, P. R.  China. She has been a professor of Department ofComputer Science, The Research  and Advanced Studies Centre of the NationalPolytechnic Institute  (CINVESTAV-IPN), Mexico. She was a senior research fellowof School of  Electronics, Electrical Engineering & Computer Science,Queen's University  Belfast, UK during the school year 2006-2007 (sabbaticalleave); and school of  Engineering, University of California Santa Cruz in 2010(sabbatical leave).  Currently she is a senior member of IEEE, member of AMC(Mexican Association of  Science), and member of SNI (National ResearcherSystem) level 2. Dr. Li has  published more than 100 papers on internationaljournals, book chapters and  conferences. She has successfully finished threeCONACYT (NSF in Mexico) projects  in the field of Knowledge and DataEngineering, and one collaborative project  with University of California Riverside.

摘要:

Support Vector Machines (SVM) has demonstrated  highly competitive performance in many real-world applications. However, despite  its good theoretical foundations and generalization performance, SVM is not  suitable for classifying large data sets because of high training complexity. In  recent years, we have introduced several data reduction techniques into SVM  classification process to handle this problem, such as minimum enclosure ball,  clustering, convex-concave hull, decision tree, etc. Experiments showed that SVM  is still suitable for large data classification if the training data is  sufficient representative. In this talk, I will present several data selection  or reduction techniques and correspondingtraining process.