On the Optimization of Search Heuristics: a Fuzzy Approach
F. Gozzo · 2005
Heuristics are key factors in reducing the computational workload of searching decision trees. Artificial intelligence techniques utilize algorithms which are highly dependent upon the heuristic sensitivity and the processing workload required to generate the plausibility of a given node. Fuzzy set theory offers a means of providing reasonable heuristics to the search procedure in an effort to improve overall heuristic performance. A procedure was defined to generate fuzzy heuristics for an nth order state-space system. Results indicate that although the fuzzy heuristic is more sensitive than a linear-weighted test heuristic and may be beneficial for certain applications, the defuzzification algorithm required by the fuzzy approach required increased computational efforts. A brief review of Al search techniques is provided as well as an illustrative example which compares the performance of the fuzzy heuristic versus a linear-weighted test heuristic.