Штучний інтелект

Анотація: This study examines the performance of Conditional Variational Autoencoder (CVAE) in handwritten digit recognition. Using the MNIST dataset, two variants of the CVAE models — convolutional and multilevel architecture — were developed and compared. The research methodology includes comprehensive data preprocessing, architecture design, training, and thorough evaluation processes. The obtained data highlight the better performance of the convolutional model-based CVAE in achieving recognition accuracy compared to its multilayer counterpart. Evaluation metrics include analysis of original and reconstructed images, comparison of hidden layer vector distribution patterns, and visualization of loss function dynamics. In addition, the study highlights the practical implications of CVAEs in various fields, highlighting their performance in digit recognition tasks due to their inherent robustness and extraordinary generalizability.

Ключові слова: neural network, autoencoder, encoder, decoder, hidden layer, image recognition

Посилання:

1. Chollet F. Deap Learning with Python / F. Chollet – Manning Publications Co., 2021. – 504 p.
2. Shujian Yu. / Principe Understanding autoencoders with information theoretic concepts / Yu Shujian, Jose C. Principe – Neural Networks Volume, 2019. – pp. 104 – 123.
3. Hoegh. On Learning Useful Variational Autoencoder Representations: Applications in Audio Modelling and Hearing Loss Treatment / Hoegh, Rasmus M. Th. – Technical University of Denmark, 2022. – 157 P.
4. Doersch C. / Tutorial on Variational Autoencoders / C. Doersch, C. Mellon. – 3 Jan. 2021, 23P.
5. Jiwoong Im. D. / Denoising Criterion for Variational Auto-Encoding Framework / D. Jiwoong Im, S. Ahn, R. Memisevic, Yo. Bengio – Montreal Institute for Learning Algorithms. University of Montreal, 4 Jan. 2016. – pp. 1–14.
6. Kaae Sonderby C. / Ladder Variational Autoencoders / C. Kaae Sonderby, N. Raiko, L. Maaloe, S. Kaae Sonderby, O. Winther. – Advances in neural information processing systems, 2016. – pp. 1 – 9.
7. Podolyak B. Y., Filimonova T. O. Development of an autoencoder for handwritten digit recognition. Collection of abstracts of the XX International Scientific and Practical Conference «Mathematical Software for Intelligent Systems. MSIS – 2022», O. Honchar DNU, Dnipro, November 22-24, 2023. - P. 243 - 244.
8. Podoliak B., Filimonova T. Development of a variational autoencoder for handwritten digit recognition. Information Technologies: Theory and Practice. I (VII) International Scientific and Practical Conference of Higher Education Applicants and Young Scientists «Information Technologies: Theory and Practice». Abstracts of Papers (Dnipro, March 20-22, 2024) / Ministry of Education and Science of Ukraine, National Technical University «Dnipro Polytechnic». - Dnipro: Svidler A.L., - 2024. - P. 160 - 162. How can you optimize performance for VAEs?: Website. URL:https://www.linkedin.com/advice/0/how-can-you-optimize-performance-vaes-7xj1f