In this paper, an algorithm is presented for learning concept classification rules. It is a hybrid between evolutionary computing and inductive logic programming (ILP). Given input of positive and negative examples, the algorithm constructs a logic program to classify these examples. The algorithm has several attractive features including the ability to explicitly use background (user-supplied) knowledge and to produce comprehensible output. We present results of using the algorithm to tackle the chess-endgame problem (KRK). The results show that using fitness proportionate selection to bias the population of ILP learners does not significantly increase classification accuracy. However, when rules are exchanged at intermediate stages in learning, in a manner similar to crossover in Genetic Programming, the predictive accuracy is frequently improved.
展开▼
