Probabilistic fault diagnosis

Probabilistic diagnostic methods are described. This approach to diagnosis has application in systems where a number of faulty states can be tolerated. Such situations exist in certain fault-tolerant computational assemblies and also in some 'AI' systems. Particular applications of the latter type include hypothesis testing, pattern recognition and learning systems. Probabilistic diagnostic methods may also find application in systems employing computational fuzzy logic where the data is normally imprecise and a certain amount of error can be tolerated.