On-line off-line Ranking Support Vector Machine and analysis

Author

Gu, Bin ; Wang, Jiandong ; Chen, Haiyan

Author_Institution

Dept. of Comput. Sci. & Eng., Nanjing Univ. of Aeronaut. & Astronaut., Nanjing

fYear

2008

fDate

1-8 June 2008

Firstpage

1364

Lastpage

1369

Abstract

Ranking support vector machine (RSVM) learning is equivalent to solving a convex quadratic programming problem. Currently there exists some difficulties for exact online ranking learning. This paper presents an exact and effective method that can solve the online ranking learning problem and shows the feasibility and finite convergence of the algorithm from the perspective of theoretical analysis. Additionally, this paper extends this method for online learning to offline ranking learning and offers another algorithm for solving large-scale RSVM.

Keywords

convex programming; learning (artificial intelligence); support vector machines; convex quadratic programming problem; finite algorithm convergence; online offline ranking support vector machine learning; Algorithm design and analysis; Approximation algorithms; Convergence; Function approximation; Machine learning; Management training; Quadratic programming; Statistical learning; Support vector machine classification; Support vector machines;

fLanguage

English

Publisher

ieee

Conference_Titel

Neural Networks, 2008. IJCNN 2008. (IEEE World Congress on Computational Intelligence). IEEE International Joint Conference on

Conference_Location

Hong Kong

ISSN

1098-7576

Print_ISBN

978-1-4244-1820-6

Electronic_ISBN

1098-7576

Type

conf

DOI

10.1109/IJCNN.2008.4633975

Filename

4633975