engleski

Development of the model of recognition of textual data using neural networks

This research is devoted to the justification, development and software implementation of the model of textual data recognition, based on the application of neural networks with hybrid architecture ConvBiGRU. The tasks of the work were to investigate the problem of text recognition, to determine the criteria for evaluating the quality of recognition, to design the architecture of the model and software application. Also, we trained the software model for text recognition, tested the software application and performed a comparative analysis of the recognition quality on other neural networks. The quality of the developed model was evaluated using the known validity metric. To compensate for possible imbalances in the samples, accuracy and completeness metrics were introduced in addition to the chosen metric. As an integrated development environment, the free cross-platform interactive environment for scientific calculations Spyder was used, running on the Anaconda platform.

textual data recognition ; neural networks ; ConvBiGRU

The conference was canceled due the low number of presenters registered for the conference, but the Book of abstracts has been published.