Synthesis of an Antecedent of the Productional Rule by Logical Neural Networks on a Basis of Architecture Similar of Group Method of Data Handling

摘要

Nowadays the requirements for the speediness of decisions are increasing in the context of the need to take into account a large number of factors characterizing the emerging situations. This is especially important in the digital reality of such areas as: sociology, medicine, ecology, etc. This makes it necessary to develop automated decision support systems based on existing or newly developed innovations of modern methodological capabilities and existing tools of artificial intelligence. Existing achievements of the Group Method of Data Handling (GMDH) allow to resolve the contradiction between a large number of informative features and a small training sample when developing the synthesis architecture of an artificial logical neural network, using the architecture of the GMDH as the basic structure and operating principles. Two types of neurons are considered: logic neuron and logic-arithmetic neuron (structures, training algorithms, recommendations for application are given). The configuration of the proposed artificial logical neural network as a whole is described. Presented are the results of experimental studies that prove the effectiveness of the proposed neural network for the synthesis of the antecedent of the production rule by the example of assessing the risk of death from thromboembolism in endoprosthetics.