A new P system with hybrid MDE-

Clustering is an important part of data mining. It can immensely simplify data complexity and helps discover the underlying patterns and knowledge from massive quantities data points. The popular efficient clustering algorithm k -means has been widely used in many fields. However, The k -means method also suffers from several drawbacks. It selects the initial cluster centers randomly that greatly influences the clustering performance. This study proposes a new P system with modified differential evolution k -means algorithm to improve the quality of initial cluster centers of k -means algorithm. The P system has three types of membranes: elementary, local store, global store. Different membranes have different rules. Based on the membrane structure, the elementary membranes evolve the objects with modified differential evolution algorithm and other types of membrane update the local best and the global best objects synchronously with communication rules. Under the control of the P system, the hybrid algorithm achieves a good partition for data sets, compared with the classical k -means algorithm and DEk -means algorithm. Key-Words: Data mining; clustering; unsupervised learning; k-means; modified DE algorithm; membrane computing.