Decision Tree Models Induced by Membrane Systems

Abstract

This paper focuses on an application of membrane systems to solve classification problems. Decision tree technique has been widely used to construct classification models because such models can closely resemble human reasoning and are easy to understand. A novel membrane computing-based decision tree induction algorithm is developed in this paper. An extended tissue membrane system with tree-like objects is considered as the computing framework, in which each object in cells expresses a feasible decision tree and the transformation-communication mechanism is applied to deal with the tree-like objects. The extended tissue membrane system with tree-like objects can efficiently induce a best decision tree model for a given data set. The proposed decision tree induction algorithm is evaluated on some data sets and compared with two classical methods.