KBG: a knowledge based generalizer
Proceedings of the seventh international conference (1990) on Machine learning
The Utility of Knowledge in Inductive Learning
Machine Learning
Elements of machine learning
Learning Logical Definitions from Relations
Machine Learning
Learning relational cliches with contextual generalization
Learning relational cliches with contextual generalization
Hi-index | 0.00 |
Top-down learners suffer often from the plateau problem (or myopia) of their greedy search algorithms. One way to address this is to extend the topdown greedy search, which grows the clauses, with relational clichÉs. Using clichÉs the search is no longer constrained to adding one literal at a time: combinations of literals instantiating clichÉs are tried as well. The paper presents CLUSE: ClichÉs Learned and USEd, a system that learns clichÉs that are then used either within a domain, or across domains. CLUSE is a bottom-up learner, in which generalization proceeds according to Contextual LGG (CLGG). GLGG is an extension of LGG that takes into account the context in which a pair of literals is generalized. The paper defines CLGG, illustrates how clichÉs are learned, and shows that the complexity of this learning is polynomial.