Unsupervised feature selection based on maximum information and minimum redundancy for hyperspectral images. (March 2016)

• Record Type: Journal Article
• Authors: Feng, Jie; Jiao, Licheng; Liu, Fang; Sun, Tao; Zhang, Xiangrong
• Is Part Of: Pattern recognition. Volume 51(2016:Mar.)
• Page Start: 295
• Page End: 309
• Publication Date: 2016-03
• Subjects: Unsupervised feature selection -- Hyperspectral images -- Maximum information and minimum redundancy -- Information theory -- Clonal selection algorithm; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2015.08.018
• Languages: English
• ISSNs: 0031-3203

Abstract

Abstract: Unsupervised feature selection plays an important role in hyperspectral image processing. It is a very challenge issue to select an effective feature subset with the unavailability of class labels. To select the features maximally preserving the information of original features, a maximum joint mutual information (MJMI) criterion is defined. Since the high-order distribution involved in MJMI is hard to calculate, a maximum information and minimum redundancy (MIMR) criterion is derived as the low-order approximation of MJMI. From information theory, many classical unsupervised feature selection criteria can also be considered as the low-order approximations of MJMI. Compared with them, MIMR requires more relaxed approximation condition. Moreover, a new clonal selection algorithm (CSA) in artificial immune system is devised to optimize the selected features with the guidance of MIMR. Experimental results on several hyperspectral datasets demonstrate that the proposed method obtains better feature subsets compared with classical unsupervised feature selection methods. Highlights: A new MIMR criterion is proposed for unsupervised feature selection. MIMR can select more informative and distinctive features. Many classical criteria and MIMR can be unified into the same framework. The theoretical advantage of MIMR over many classical criteria is given. Feature selection problem is solved by combinatorial optimization with a new CSA.