Robust Method for Training Sequential Decision-Making for Artificial Intelligence Systems
Original Publication Date: 1988-Jun-01
Included in the Prior Art Database: 2005-Feb-15
A technique is described whereby a robust method is implemented in the training process of sequential decision making, as associated with artificial intelligence applications. Algorithms are utilized that provide a method of determining an optimal ordering of search alternatives in a chain of decisions, such that faster execution is achieved. The concept is an improvement over prior art concepts which utilized adaptive search techniques. In prior art, the idea of ordering a set of search alternatives to improve search performance showed that the optimal ordering is dependent on the probability of success and the cost of evaluating the different alternatives. Algorithms have been devised for adaptively organizing a sequential search, where the costs of individual outcomes are equal.