A new machine learning system using concept theory based rule induction

A machine learning system through rule induction by the application of concept theory (CT) is described. Conceptual clustering of examples (CCE), one form of learning by discovery, determines the natural classes of input training instances. A concept for each class is thereafter formed. CT is then applied over those concepts to induce concept rules (CR). Further modifications of CR are performed by the application of the technique of learning by analogy to the input instances and CR, thereby inducing modified concept rules (MCR).<
>