Explicit Representation of Concept Negation

摘要

We present a learning method called Negative Explanation Based Generalization (NEBG) that performs automatic changes of representation by computing the negation of an already known concept. NEBG is similar to EBG as a deductive and valid learning method using a single example. It is based on new logic programming techniques based on example-guided transformation of the completed database. We also introduce a very powerful heuristic based on functional properties of the application domain. The implemented algorithms are described and several examples are given.