Adaptive hierarchical censored production rule-based system: A genetic algorithm approach

Abstract

An adaptive system called GBHCPR (Genetic Based Hierarchical Censored Production Rule) system based on Hierarchical Censored Production Rule (HCPR) system is presented that relies on development of some ties between Genetic Based Machine Learning (GBML) and symbolic machine learning. Several genetic operators are suggested that include advanced genetic operators, namely, Fusion and Fission. An appropriate credit apportionment scheme is developed that supports both forwardand backward chaining of reasoning process. A scheme for credit revision during the operationsof the genetic operators Fusion and Fission is also presented. A prototype implementation is included and experimental results are presented to demonstrate the performance of the proposed system.