Neuromorphic spiking neural network-oriented electronic afferent neuron and implementation method thereof

Abstract

The invention relates to an electronic afferent neuron in the field of brain-like intelligence, and solves the technical problems that an electronic afferent neuron constructed by a traditional CMOSFET is high in power consumption, complex in circuit structure and process, difficult to apply to a novel neuromorphic spiking neural network and the like. The invention provides an electronic afferentneuron oriented to a neuromorphic spiking neural network. A metal-insulator conversion MIT memristor RM, a multi-input-end floating gate MIFG transistor T0, a multi-resistance-state memristor RD and acapacitor CF are adopted, conversion from an analog signal to a pulse signal is achieved, and the memristor is an interface of a sensor and a neuromorphic pulse neural network. Meanwhile, the invention is based on the electronic afferent neurons oriented to the neuromorphic spiking neural network. The invention further provides a neuromorphic spiking neural network-oriented electronic afferent neuron implementation method, information integration and frequency coding are performed on input biological neurons, and programmable threshold output is realized by programming control of an output pulse frequency threshold.

Description

technical field [0001] The invention relates to an electronic afferent neuron in the field of brain-inspired intelligence, in particular to an electronic afferent neuron oriented to a neuromorphic impulse neural network and an implementation method thereof. Background technique [0002] Artificial intelligence has become the core driving force of a new round of technological revolution, and brain-inspired intelligence technology will lead the development direction of artificial intelligence in the future. With the rapid integration and development of neuromorphic computing (Neuromorphic Computing, NC) and bionic sensing technology, brain-like intelligent robots have taken shape. On the information processing side, by simulating the information processing method of "computing-storage integration" of the human brain, the performance of NC Spiking Neural Network (SNN) has been rapidly improved. Under the condition that the system power consumption is much lower than that of th...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the technical problems of high power consumption, complex circuit structure and process, and difficulty in applying to new neuromorphic impulse neural networks in the electronic input neurons constructed by traditional CMOSFETs, and propose a neuromorphic impulse neural network-oriented Electron afferent neuron and its realization method

Images