MULTISCALE MODELING OF NEUROMORPHIC SYSTEMS

Multiscale approaches, models and algorithms for designing neuromorphic devices for the memory of new generation computers are presented. The developed approaches make it possible to solve problems associated with simulating modeling of the work of neuromorphic networks in intellectual analysis modes of data and machine learning. The issues related to the construction of a computing model of formation/destruction of conductive channels in memristive elements underlying neuromorphic networks are considered. New algorithms for modeling the work of the neuromorphic network have been created, taking into account stochastic effects, as well as original methods and means of simulating modeling of teaching a neuromorphic network. The main approaches are presented when creating software for simulating modeling of neuromorphic networks based on the integration platform of multiscale modeling, combining information flows at various large -scale levels, including the level of a new resistive memory element, the level of neuromorphic network and the level of simulation of the neuromorphic network on precedents . The book is intended for scientists, specialists in the field of computing electronics, senior students and graduate students of technical universities.