СОВРЕМЕННЫЕ ПРОБЛЕМЫ КОМПЬЮТЕРНЫХ И ИНФОРМАЦИОННЫХ НАУК, VI Международная научная конференция «Конвергентные когнитивно-информационные технологии»

Размер шрифта: 
Convolutional Artificial Immune System
Rostislav Mikhailovich Mikherskii, Mikhail Rostislavovich Mikherskii

Изменена: 2021-11-03

Реферат


A new machine learning algorithm is proposed - a convolutional artificial immune system. Based on this algorithm, a text classification system is implemented programmatically. This system has demonstrated high speed of work and training, as well as good quality classification of texts. As an example of the work of this system, the classification of texts by their two authors was considered: Russian writers Mikhail Sholokhov and FedorKryukov. For the training of the system, 10 works of each of the authors written in Russian were selected. The system training time on a laptop with an Intel Core i5-3210M processor was 233.819 s. To check the quality of text recognition, 6 works of each of the authors that were not used in the training of the system were selected. The system made no mistakes in recognizing these texts. In addition, a study was conducted of the text of the novel "The Quiet Don", with the aim of resolving the question of which of these writers is its author. It is shown that FedorKryukov cannot be the author of the novel "The Quiet Don". It is very likely that the writer Mikhail Sholokhov is indeed the author of this novel.