Utilizing Prior Concepts for Learning

The inductive learning problem consists of learning a concept given examples and nonexamples of the concept. To perform this learning task, inductive learning algorithms bias their learning method. Here we discuss biasing the learning method to use previously learned concepts from the same domain. These learned concepts highlight useful information for other concepts in the domain. We describe a transference bias and present M-FOCL, a Horn clause relational learning algorithm, that utilizes this bias to learn multiple concepts. We provide preliminary empirical evaluation to show the e ects of biasing previous information on noise-free and noisy data.