An improved spectral clustering algorithm based on dynamic tissue-like membrane system

Abstract

With vast amount of data generated, it is becoming a main aspect to mine useful information from such data. Clustering research is an important task of data mining. Traditional clustering algorithms such as K-means algorithm are too old to propose high-dimensional data, so an efficient clustering algorithm, spectral clustering is generated. In recent years, more and more scholars has been firmly committing to studying spectral clustering algorithm for its solid theoretical foundation and excellent clustering results. In this paper we propose an improved spectral clustering algorithm based on Dynamic Tissue-like P System abbreviated as ISC-DTP. ISC-DTP algorithm takes use of the advantages of maximal parallelism in tissue-like membrane system. Experiment is conducted on an artificial data set and four UCI data sets. And we compare the ISC-DTP algorithm with original spectral clustering algorithm and K-means algorithm. The experiments demonstrate the effectiveness and robustness of the proposed algorithm.