This paper is about the method of pattern recognition using the example of handwritten letters of the English alphabet. A program was developed to solve this task. This program uses recognition algorithms based on neural networks. The paper studies several variants of the architecture of neural networks and their parameters. To train and test the effectiveness of the program, an open data set EMNIST (an extension of MNIST to handwritten letters) is used. The results of letters recognition on specific examples are demonstrated.
eLIBRARY: https://www.elibrary.ru/item.asp?id=35359575
RUDN. Repository: https://repository.rudn.ru/en/records/article/record/14524/
Volkov S. S. The system of handwritten letters recognition based on neural network // Information and Telecommunication Technologies and Mathematical Modelling of Hi-tech Systems. - 2018. - Page 159-161.