title : Provably Fast Training Algorithms for Support Vector Machines

title: Provably Fast Training Algorithms for Support Vector Machines author: Jose Balcázar1, Yang Dai2, Junichi Tanaka3, and Osamu Watanabe3 affiliation: 1. Departament de Llenguatges i Sistemes Informatics, Univ. Politecnica de Catalunya Campus Nord, Jordi Girona Salgado 1-3, 08034 Barcelona, Spain 2. Dept. of Bioengineering (MC063), Univ. Illionis at Chicago 851 S. Morgan Str, Chicago, IL 60607-7052, USA 3. Dept. of Mathematical and Computing Sciences, Tokyo Institute of Technology Meguro-ku Ookayama, Tokyo 152-8552, Japan email: [email protected] acknowledgments to financial supports: The first and the forth authors started this research while visiting the Centre de Recerca Mathemàtica of the Institute of catalan Studies in Barcelona. The first author is supported by IST Programme of the EU under contract number IST-199914186 (ALCOM-FT), EU EP27150 (Neurocolt II), Spanish Government PB98-0937C04 (FRESCO), and CIRIT 1997SGR-00366. The second author conducted this research while she is with Dept. of Mathematical and Computing Sciences, Tokyo Institue of Technology, and is supported by a Grantin-Aid (C-13650444) from Japanese Goverment. The forth author is supported in part by a Grant-in-Aid for Scientific Research on Priority Areas “Discovery Science” from Japanese Goverment.