INDUCTION ALGORITHMS: GUEST EDITOR'S INTRODUCTION

摘要: Induction algorithms are becoming more and more important in the production of commercial expert systems since automatic rule generation from examples not only saves time, but is quite often the only means of extracting knowledge. They are particularly important in areas where human experts are not available, or where obtaining rules from any other sources is impossible. In some situations there are no known rules. For example, in a copper refinery process (controlled by 35 variables) there must be a decision made as to adding one of two agents: glue or tembine. It is known, from observation, that adding the agent may improve the quality of the copper, but there is no theory explaining this phenomenon. The knowledge leading to rules of adding the agent must be obtained, in this case, by learning from examples.Induction algorithms are essential components of expert systems coping with bulk data. Their …