Optimization of Ensemble based Decision using PSO

In matter of great importance, it is the innate behaviour for human beings to seek more than one consultant before making a decision. In such cases we weigh the individual opinions of experts on the basis of their competence and then combine them through some thought process in order to finalize the decision. In this paper we have proposed an idea of Particle Swarm Optimization (PSO) in order to optimize these weights which then better evaluate the competence of an expert. Weighted Majority Voting (WMV) is the most popular technique used to combine such opinions in an ensemble based classification. The weights associated to each base classifier in WMV on the basis of its competence are optimized under the influence of the basic idea of PSO. PSO has shown the stable performance on the selected datasets from UCI Repository and generally improved the performance of an ensemble system.