The version-space algorithm is a form of concept learning. Version spaces attempt to create a ruke to identify a single class form posituve and negative examples. It keepts track of the most general and most specific rules that are consistent with all examples seen so far until there is either one rule left (succes) or none (failure!). The latter makes it sensitive to noisy input.
Defined on page 94
Used on Chap. 5: pages 92, 94, 107; Chap. 9: page 176
Also known as version spaces, version-space
Rule lattice