Experiments in the Automatic Selection of Problem-solving Strategies
Matthias Fuchs · 1999
We present an approach to automating the selection of search-guiding heuristics that control the search conducted by a problem solver. The approach centers on representing problems with feature vectors that are vectors of numerical values. Thus, similarity between problems can be determined by using a distance measure on feature vectors. Given a database of problems, each problem being associated with the heuristic that was used to solve it, heuristics to be employed to solve a novel problem are suggested in correspondence with the similarity between the novel problem and problems of the database. Our approach is strongly connected with instance-based learning and nearestneighbor classification and therefore possesses incremental learning capabilities. In experimental studies it has proven to be a viable tool for achieving the final and crucial missing piece of automation of problem solving---namely selecting an appropriate search-guiding heuristic---in a flexible way. This work was ...