Deep neural network system based on memristor

Abstract

The invention provides a deep neural network system based on a memristor, which comprises an input layer, an output layer and multiple hidden layers, wherein the input layer receives an external information input mode, the external information input mode is calculated and converted layer by layer, and an external output result is generated finally by the output layer. A synaptic weight of the deep neural network adopts the memristor for simulation, and the feature that resistance of the memristor changes along with applied electric signals is used to simulate a connection strength behavior of a connection synapse between neural networks. The invention further provides an information processing system for the deep neural network system based on the memristor.

Description

technical field [0001] The invention relates to a deep neural network system. Background technique [0002] In 2006, Geoffrey Hinton, a professor at the University of Toronto in Canada, published a paper "Reducing the Dimensionality of Data with Neural Networks" in "Science", which set off a new wave of artificial neural network research based on deep neural networks. Deep Neural Networks (DNN) are different from traditional artificial neural networks. They mainly draw on the characteristics of biological neural networks with multi-layer topology, and artificially construct artificial neural networks with multiple hidden layers. Due to the combination of the greedy unsupervised algorithm of "layer-by-layer pre-training" and the global adjustment algorithm, this multi-layer neural network is relatively easy to converge. Compared with shallow neural network models, deep neural networks have outstanding feature learning capabilities, and the learned features can express data m...

AI Technical Summary

Problems solved by technology

The above calculation process is based on the von Neumann computer serial paradigm in the traditional artificial neural network, resulting in huge overheads in terms of size, energy consumption, and time, making it difficult to apply in the embedded field

Images