Open AccessParameter optimization of support vector machine based on improved grid algorithm
Author(s) -
Shenghua Wang,
Li Dong,
Huichun Hua
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1693/1/012108
Subject(s) - hyperplane , support vector machine , computer science , algorithm , construct (python library) , grid , range (aeronautics) , sequential minimal optimization , artificial intelligence , machine learning , mathematics , engineering , geometry , programming language , aerospace engineering
Support vector machines SVM) is a very popular algorithm used widely in the classification problem. The basic idea of SVM is to construct two parallel hyperplanes to separate two classes of instances and maximize the distance between the hyperplanes. In this paper, we propose a new algorithm called SVM-ICC-DBTCPRP one to solve the parameter optimization problem by improving the grid algorithm used by Fayed and Atiya. In the new algorithm there are two main sub-algorithms that are used to preselect the support vectors for reducing the time of training and preselect the parameter range to reduce the number of training respectively. Six typical data sets are selected to verify the effectiveness of our algorithm. The computed results show that our algorithm has the obvious advantage on the aspect of elapsed time than the one of Fayed and Atiya.