Paper Contents
Abstract
The study of strategies for picking up broad concepts or rules from a set of training instances is a key area of research in artificial intelligence. A guaranteed method is provided for solving this problem that finds all rule versions that are consistent with a collection of positive and negative instances for training without going backtracking. The proposed algorithm makes use of a representation of that rule space that is in agreement with the observed training data. Candidate rule versions that are discovered to clash with each new instance are removed from this "rule version space" in response to new training instances. The application of version spaces is explained in relation to Meta-DENDRAL, a software that acquires knowledge about chemical spectroscopy rules.
Copyright
Copyright © 2024 DHARSHINI . K. This is an open access article distributed under the Creative Commons Attribution License.
