A study of the accuracy of heuristic functions
Carlos Linares López · European Conference on Artificial Intelligence · 2004
From a practical point of view, most problems cannot be optimally solved applying brute-force single-agent search algorithms. Thus, a new class of single-agent search algorithms was devised to find the optimal solution faster and/or consuming less space, heuristic single-agent search algorithms. However, designing accurate heuristic functions is still a major challenge. In this paper, a mathematical model is introduced for characterizing the accuracy of any heuristic function. This is a step ahead in the analysis of heuristic functions since this method could be used to aid other procedures to improve the original heuristic estimations.