Artificial Intelligence
Probabilistic Inference and Baysian Theorem Based on Logical Implication
RSFDGrC '99 Proceedings of the 7th International Workshop on New Directions in Rough Sets, Data Mining, and Granular-Soft Computing
An overview of conditionals and biconditionals in probability
MATH'08 Proceedings of the American Conference on Applied Mathematics
Fuzzy Sets and Systems
Fuzzy Sets and Systems
Hi-index | 0.00 |
A classical method for ranking n potential events as sources of error is Bayes' theorem. However, a ranking based on Bayes' theorem lacks a fundamental symmetry: the ranking in terms of blame for error will not be the reverse of the ranking in terms of credit for lack of error. While this is not a flaw in Bayes' theorem, it does lead one to inquire whether there are related methods which have such symmetry. Related methods explored here include the logical version of Bayes' theorem based on probabilities of conditionals, probabilities of biconditionals, and ratios or differences of credit to blame. We find that of all the methods described, probabilities of biconditionals and a corresponding notion of logical correlation coefficients provide a particularly attractive method for ranking blame for error and credit for lack of error which has the symmetry property we are interested in.