Open AccessRESEARCH OF GENERATIVE ADVERSARIAL NETWORKS FOR THE SYNTHESIS OF NEW MEDICAL DATA
Author(s) -
V Laptev,
Вячеслав Данилов,
Olga Gerget
Publication year - 2020
Publication title -
avtomatizaciâ i modelirovanie v proektirovanii i upravlenii
Language(s) - English
Resource type - Journals
eISSN - 2658-6436
pISSN - 2658-3488
DOI - 10.30987/2658-6436-2020-2-17-23
Subject(s) - discriminator , generator (circuit theory) , computer science , task (project management) , artificial intelligence , generative grammar , generative adversarial network , artificial neural network , sample (material) , machine learning , pattern recognition (psychology) , deep learning , engineering , telecommunications , power (physics) , chemistry , physics , systems engineering , chromatography , quantum mechanics , detector
The paper considers the development of a Generative Adversarial Network (GAN) for the synthesis of new medical data. The developed GAN consists of two models trained simultaneously: a generative model (G - Generator), estimating the distribution of data, and a discriminating model (D - Discriminator), which estimates the probability that the sample is obtained from the training data, and not from generator G. To create G, we used own neural network architecture based on convolutional layers using experimental functions of Tensor Flow Addons. To create discriminator D, we used a Transfer Learning (TL) approach. The training procedure is to maximize the likelihood that discriminator D will make a mistake. Experiments show that the proposed GAN architecture completely copes with the task of synthesizing of new medical data.