APPROXIMATE REPRESENTATION OF PROBABILISTIC DATA IN EXPERT-SYSTEMS

We propose the use of a third-order approximation for the representati on of probabilistic data in expert systems and compare it to tree-stru ctured representations. The differences are illustrated using the exam ple of a reliability problem. We show that using the third-order repre sentation results in significantly reduced losses as compared to tree structures, with a small increase in computational complexity, We pres ent heuristic and exact techniques to determine the optimal third-orde r representation and propose a decomposition technique that allows the exact algorithm to be efficiently used for solving large problem inst ances.