Feature selection for multi-class classification using support vector data description
Daun Jeong, Dongyeop Kang, Sangchul Won · 2010
In this paper, a supervised feature selection approach is presented, which is based on support vector data description(SVDD). This method is suggested for multi-class classification case, and it utilizes a sequential backward selection algorithm using the accuracy of classifier to decide which feature to be eliminated. The proposed approach is applied to well-known real world datasets, and the obtained results are compared with results from the existing feature selection techniques. Simulation results demonstrate the effectiveness of the proposed method.