Resumen
Over the past few years, deep learning methods have made significant progress and have become widespread tools for solving various cognitive tasks. To leverage the power of deep learning everywhere, you need to deploy deep learning not only to large-scale computers but also to peripherals. However, the ever-growing complexity of deep neural networks, coupled with a dramatic increase in the amount of data processed, place significant energy demands on modern computing platforms. The neuromorphic computing model assumes that computations are performed in a biologically plausible way. Part of neuromorphic computing is spike neural networks, which are one of the leading candidates for overcoming the limitations of neural computing and effectively using machine learning algorithms in real-world applications. This paper discusses the biological foundations of spike neural networks, methods for training and creating them, as well as software and hardware platforms for their use.