Casa Robino

One way to assess the suitability of a refinement operator for learning algorithms is to look at its properties. We analyse the properties (completeness, weak completeness, properness, redundancy, finiteness, minimality, and their combinations) and show theoretical limitations.<br />
<br />
Learning algorithms can be designed by combining a refinement operator with a a search heuristic. We are interested in designing operators which are close to the best one can hope for and combine them with intelligent search heuristics.