Methods of effective data analysis for solving poorly formalized recognition tasks

The paper presents a comparative analysis of the results of neural network solutions and logical algorithms. A method is proposed to improve the results of neural network solutions. The work considers a sigma-pi neural network as a solver, and the data are represented by objects and features characterizing these objects. The paper shows that there are logical approaches that can most fully describe all possible patterns in the study area. At the same time, neural networks provide only a part of the possible solutions as solutions. Therefore, it is proposed to analyze and supplement these solutions. To do this, the paper considers the possibility of constructing a logical corrector capable of constructing a logical function according to the structure of a neural network, and then implementing it in the form of logical neural networks. This approach allows to identify logical relationships between objects in the data under study, knowledge of logical patterns will allow to formalize and more accurately understand the nature of the analyzed area.