The Community for Technology Leaders
2014 IEEE International Conference on Data Mining Workshop (ICDMW) (2014)
Shenzhen, China
Dec. 14, 2014 to Dec. 14, 2014
ISBN: 978-1-4799-4275-6
pp: 74-79
ABSTRACT
In this paper, we propose a robust proximal classifier via absolute value inequalities (AVIPC) for pattern classification. AVIPC determines K proximal planes by solving K optimization problems with absolute value inequalities. In AVIPC, each proximal plane is closer to one class and far away from the others. By using the absolute value inequalities, AVIPC is more robust and sparse than traditional proximal classifiers. The optimization problems can be solved by an iterative algorithm, and its convergence has been proved. Preliminary experimental results on visual and public available datasets show the comparable performance and stability of the proposed method.
INDEX TERMS
Support vector machines, Educational institutions, Accuracy, Electronic mail, Robustness, Optimization, Training,linear program, pattern recognition, proximal classifier, sparse learning, absolute value inequalities
CITATION
Yuan-Hai Shao, Chun-Na Li, Zhen Wang, Ming-Zeng Liu, Nai-Yang Deng, "Proximal Classifier via Absolute Value Inequalities", 2014 IEEE International Conference on Data Mining Workshop (ICDMW), vol. 00, no. , pp. 74-79, 2014, doi:10.1109/ICDMW.2014.14
88 ms
(Ver 3.3 (11022016))