Fuzzy subspace clustering algorithm based on modified firefly algorithm

Aiming at the clustering problems in selected features and the shortcomings that fuzzy C-means clustering is sensitive to initial value and easy to fall into local optimum, a new fuzzy subspace clustering algorithm based on improved firefly algorithms is proposed. Based on fuzzy C-means clustering algorithm, the algorithm uses a method of calculating feature weighting in reliability-based k-means algorithm, and combines with the global search capability of firefly algorithm to search for all the subspace. An objective function is designed to evaluate the clustering results and feature-dimension included in subspace, and it is used to improve the search formula of firefly algorithm. Experimental results show that the proposed algorithm can effectively converge to the global optimal solution, and has good clustering effect and noise immunity.