Feature-reduction fuzzy co-clustering approach for hyper-spectral image analysis

摘要: Abstract Fuzzy co-clustering algorithms are the effective techniques for multi-dimensional clustering in which all features considered of equal importance (relevance). In fact, features’ could be different, even several them redundant. The removal redundant has formed idea feature-reduction problems big data processing. this paper, we propose a new unsupervised learning scheme by incorporating feature-weighted entropy into objective function fuzzy co-clustering, called Feature-Reduction Co-Clustering Algorithm (FRFCoC). First, is on basis original adds parameters representing weight different features. Next, and automatic schema adjusted based FCoC’s calculates conditions to eliminate irrelevant feature components. FRFCoC algorithm can mathematically shown converge after finite number iterations. experiment results were conducted some many-features sets hyperspectral images that have demonstrated outstanding performance compared with previously proposed algorithms.