Evaluation of Uncertain Inference Models I: PROSPECTOR
This work assesses a specific uncertain inference model for researchers in AI reasoning, identifying its limitations as an incremental analysis.
The paper evaluated the accuracy of the PROSPECTOR uncertain reasoning model by comparing its solutions to probability theory and minimum cross-entropy calculations on computer-generated inference networks. It found that PROSPECTOR was generally accurate only for a restricted subset of problems consistent with its assumptions, but even there, performance deteriorated under certain conditions.
This paper examines the accuracy of the PROSPECTOR model for uncertain reasoning. PROSPECTOR's solutions for a large number of computer-generated inference networks were compared to those obtained from probability theory and minimum cross-entropy calculations. PROSPECTOR's answers were generally accurate for a restricted subset of problems that are consistent with its assumptions. However, even within this subset, we identified conditions under which PROSPECTOR's performance deteriorates.