Improved K-Medoids Clustering Algorithm for a Class of P Systems

This paper presents an improved K-medoids clustering algorithm, which provides a new idea and method for clustering analysis[1]. The P membrane system has great parallielism[2]. This way can reduce the computing time and increase the computing efficiency, and is especially suitable for solving such problems like clustering. First, this paper proposes an improved algorithm based on the K-medoids algorithm[3]. This algorithm is based on the improved K-nearest algorithm. The initial point of the cluster is the K specific points that are selected in a specific way to reduce the number of operations, rather than from all non-central objects randomly selected in the original algorithm. Secondly, the paper P system has built all of the rules to solve the problem of K-medoids. The manner in which a particular P system is designed to calculate distances associated with n objects, such as a distance matrix or the like. The calculation of the system can lead to a possible classification in a non-deterministic way. Through multiple algorithm tests, the algorithm is suitable for cluster analysis and optimizes the K-medoids algorithm. This is a new attempt at membrane system applications.