Ключевое слово: «neural networks»

In this work, a theoretical basis will be given about neural networks, as well as their potential. In its practical part, we will create our own neural network for a brighter idea of this technology. The work will be done using a high-level programming language – Python. Assistive tools in this case will be the activator function – sigmoid, as well as the back propagation method for aligning synaptic weights. As a result, we have successfully created NN, which is able to predict the output data from the input data. The developed NN can be considered as a convenient tool for practical understanding of how neural networks work.

Machine Translation (MT) has evolved from rigid rule-based systems to statistical methods and finally to neural architectures. While Supervised Neural Machine Translation (NMT) achieves high performance, it depends heavily on large parallel corpora. This survey reviews the techni- cal evolution of MT state-of-the-art (SOTA) approaches and analyzes Unsupervised Machine Translation (UMT). We examine how modern models learn translation mappings using only monolingual data, de- coupling performance from the availability of bilingual datasets.