THE REFINEMENT OF PROBABILISTIC RULE SETS - SOCIOPATHIC INTERACTIONS

Probabilistic rules in a classification expert system can result in a sociopathic knowledge base, as a consequence of the assumption of cond itional independence between observations and rule modularity. A socio pathic knowledge base has the property that all the rules are individu ally judged to be correct rules, yet a subset of the knowledge base gi ves better classification accuracy than the original knowledge base, i ndependent of the amount of computational resources that are available . This paper describes how sociopathic interactions cause rule inducti on and refinement methods to converge to local optima with respect to maximizing classification accuracy. The problem of optimally refining sociopathic knowledge bases is modeled as a bipartite graph minimizati on problem and shown to be NP-hard. A heuristic rule refinement algori thm for sociopathic reduction, called SOCIO-REDUCER, is presented. Exp erimental results in a medical diagnosis domain show that it can reduc e the diagnosis error rate by 10.5%.