Journal of Mathematical Sciences: Advances and ApplicationsAuthor's: Cheng-Huei Yang, Li-Yeh Chuang and Cheng-Hong Yang
Pages: [199] - [227]
Received Date: May 6, 2009
Submitted by:
The feature selection process can be considered a problem of global combinatorial optimization in machine learning, which reduces the number of features, removes irrelevant, noisy and redundant data, and results in acceptable classification accuracy. Feature selection is of great importance in the fields of data analysis and information retrieval processing, pattern classification, and data mining applications. In this paper, we propose to combine tabu search (TS) and improved binary particle swarm optimization (IBPSO) for feature selection. IBPSO serves as a local optimizer each time TS has been executed for a single generation. The K-nearest neighbor (K-NN) method with leave-one-out cross-validation (LOOCV) serves as an evaluator of the TS and IBPSO procedures. The proposed method is applied and compared to ten data classification problems taken from the literature and the results are compared to other feature selection methods. Experimental results show that the proposed method simplifies features effectively and either obtains higher classification accuracy or uses fewer features compared to other feature selection methods.
feature selection, tabu search, improved particle swarm optimization, K-nearest neighbor, leave-one-out cross-validation.
