Abstract:
A method for constructing logical neural networks based on variable-valued logic functions is proposed. A theorem on the possibility of representing any logical function as a logical neural network is proved. The proof also contains an algorithm for constructing a logical neural network. The possibility of a generalization of the result obtained to the case of fuzzy logic is indicated.
Keywords:variable-valued predicate, data mining, variable-valued logical function, training sample, neural network approach, logical neural network, fuzzy logic.
Fulltext:
PDF file (2022 kB)