A parellel genetic algorithm approach for automated discovery of censored production rules

Exceptions, which focus on a very small portion of dataset, have been discarded as noise in machine learning. It is interesting to discover exceptions, as they challenge the existing knowledge and often lead to the growth of knowledge in new directions. Discovering exceptions from voluminous data sets still remains a great challenge. In Knowledge Discovery in Databases (KDD), it is also significant to extract the knowledge in a form that is flexible and efficient enough for approximate reasoning. A Censored Production Rule (CPR) is a special kind of knowledge structure that is represented as an augmented production rule of the form: If P Then D Unless C, where C (Censor) is an exception to rule. If P Then D part of CPR holds frequently while Unless C part holds rarely. Discovery of CPRs not only provides a computational mechanism to exhibit variable precision logic but also results in the discovery of a set of rules of considerably reduced size. In this paper, a Parallel Genetic Algorithm approach is suggested for automated discovery of Censored Production Rules. The parallel Genetic Algorithm involves both data and control parallelism. A fitness function that incorporates the constraints of CPRs, is proposed. The experimental results establish the effectiveness of the proposed algorithm.