Published June 30, 2022 | Version v1
Journal article Open

Use of Multilayer LSTM Neural Network in the Process of Printed Texts Recognition

Creators

  • 1. State University of Infrastructure and Technology, Ukraine

Description

The purpose of the article is to study, analyze and consider general problems and prospects for the development of printed text recognition systems based on the use of neural networks.

The research methodology consists in methods of semantic analysis of this subject area’s basic concepts (recognition systems of printed texts). Approaches to the development and operation of recognition systems based on neural networks are considered.

The scientific novelty of the research is the development of its own approach to text recognition based on neural networks, the results of which were used in the development of its own system of print recognition.

Conclusions. The paper considers the well-known views on pattern recognition on the example of printed texts and analyzes modern approaches to the use of neural networks and their training. Taking into account the results of the analysis, the authors decided to develop a system for recognizing the languages of printed texts using learning neural networks.

Files

Use_of_Multilayer_LSTM_Neural_Network_in_the_Process_of_Printed_Texts_Recognition.pdf

Files (677.0 kB)

Additional details

References

  • Alekseev, P.P. and Kviatkovskaia, I.Iu., 2021. Primenenie neironnykh setei dlia raspoznavaniia printcipialnykh uslovno-graficheskikh elektricheskikh oboznachenii [The use of neural networks for the recognition of fundamental conventional graphic electrical symbols]. Vestnik Astrakhanskogo gosudarstvennogo tekhnicheskogo universiteta. Seriia: Upravlenie, vychislitelnaia tekhnika i informatika, [online] 2, pp.47-56. Available at: <http://www.mathnet.ru/links/57f0ef0c3d69c2f4645481ceae10f0f4/vagtu669.pdf> [Accessed 25 April 2022].
  • Bengio, Y., 2009. Learning deep architectures for AI. Foundations and Trends in Machine Learning, 2, pp.1-127.
  • Bengio, Y., LeCun, Y. and Hinton G. 2015. Deep Learning. Nature, 521. pp.436-444.
  • Burlieiev, O., Vasylenko, O. and Ivanenko, R., 2021. Efektyvnist vykorystannia shtuchnykh neironnykh merezh v ekonomitsi [Efficiency of use of artificial neural networks in economy]. Ekonomika ta suspilstvo, [online] 31. Available at: <https://doi.org/10.32782/2524-0072/2021-31-27> [Accessed 22 April 2022].
  • Gers, F., Schraudolph, N. and Schmidhuber, J. 2002. Learning precise timing with LSTM recurrent networks. Journal of Machine Learning Research, 3, pp.115-143.
  • Girshick, R., 2015. Fast R-CNN. Computer Vision and Pattern Recognition, [online] Available at: <https://arxiv.org/abs/1504.08083> [Accessed 22 April 2022].
  • Goodfellow, I.J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A. and Bengio, Y., 2014. Generative Adversarial Networks. Cornell University, [online] 10 June. Available at: <https://arxiv.org/pdf/1406.2661.pdf> [Accessed 25 April 2022].
  • Leshchynskyi, O.L. and Ishchenko, A.O., 2017. Vykorystannia neiromerezh u protsesi intelektualnoho (klasternoho) analizu danykh [The use of neural networks in the process of intelligent (cluster) data analysis]. Ekonomika ta suspilstvo, [online] 11, pp.578-581. Available at: <https://economyandsociety.in.ua/journals/11_ukr/93.pdf> [Accessed 23 April 2022].
  • Markova, S.V. and Zhigalov, K.Iu., 2017. Primenenie neironnoi seti dlia sozdaniia sistemy raspoznavaniia izobrazhenii [Application of a neural network to create an image recognition system]. Fundamentalnye issledovaniia, [online] 8 (1), pp.60-64. Available at: <http://www.fundamental-research.ru/ru/article/view?id=41621> [Accessed 25 April 2022].
  • Oleinik, A., Zybina, E. and Oleinik, E., 2020. Obzor Reshenii Neironnykh Setei Primeniaemykh dlia Raspoznavaniia Obrazov [Overview of Neural Network Solutions Applied to Pattern Recognition]. In: Informatsiini systemy ta tekhnolohii IST-2020 [Information systems and technologies IST-2020]. 9th International Scientific and Technical Conference. Kharkiv, Ukraine, [online] November 17-20, 2020, pp.128-131. Kharkiv: Madrid Printing House. Available at: <https://openarchive.nure.ua/bitstream/document/16163/1/IST-2020-128-131.pdf> [Accessed 25 April 2022].
  • Schmidhuber, J., 2015. Deep Learning in Neural Networks: An Overview. Neural Networks, 61, pp.85-117.
  • Stavytskyi, O.V. and Mozolevska, M.O., 2017. Vykorystannia neironnykh merezh dlia prohnozuvannia u finansovii sferi [The use of neural networks for forecasting in the financial sector]. Aktualni problemy ekonomiky ta upravlinnia, [online] 11. Available at: <http://ape.fmm.kpi.ua/article/view/102584> [Accessed 23 April 2022].
  • Subbotin, S.O., Oliinyk, A.O. and Oliinyk, O.O., 2009. Neiteratyvni, evoliutsiini ta multyahentni metody syntezu nechitkolohichnykh i neiromerezhnykh modelei [Non-iterative, evolutionary and multiagent methods of synthesis of fuzzy and neural network models]. Zaporozhye: ZNTU.
  • What are Recurrent Neural Networks. [online] Available at: <https://www.ibm.com/cloud/learn/recurrent-neural-networks> [Accessed 22 April 2022].
  • Zheng, Y., Meng, Y. and Jin, Y., 2011. Object Recognition using Neural Networks with Bottom-up and Top-down Pathways. Neurocomputing, 74, pp.3158-3169.