A New Embedded Feature Selection Method using IBALO mixed with MRMR criteria. (January 2020)

• Record Type: Journal Article
• Authors: Zhang, Yu; Zhao, Zhuanzhe; Zhao, Shuaishuai; Liu, Yongming; He, Kang
• Is Part Of: Journal of physics. Volume 1453(2020)
• Publication Date: 2020-01
• Subjects: Physics -- Congresses; 530.5
• DOI: 10.1088/1742-6596/1453/1/012027
• Languages: English
• ISSNs: 1742-6588

Abstract

Abstract: In order to remove irrelevant data and increase classification accuracy in feature selection, this paper proposed a new Embedded feature selection method with gathering Minimal Redundancy Maximal Relevance (MRMR) criteria, Sequential Forward Selection (SFS) and Improved Binary Ant Lion Optimizer (IBALO) together. Totally, we use three different feature selection methods namely MRMR mixed with Sequential Forward Selection (MS), MS mixed with Binary Ant Lion Optimizer (MS-BALO), an improvement of MS-BALO. Experiments prove that the MS-IBALO method proposed by this paper is efficient compared to MS and MS-BALO methods.