A kind of spiking neural network neuron circuit based on lif model

Abstract

The invention claims protection of a pulse neural network neuron circuit based on a LIF (Leaky Integrate and Fire) model, which belongs to the field of integrated circuit design. The circuit mainly includes: a membrane potential accumulation circuit, a leakage circuit, a pulse generation circuit, a refractory period circuit and a reset circuit. The invention not only realizes the integral leakage ignition function of the LIF neuron model, but also simulates the refractory period characteristic of the biological neuron, wherein the refractory period is adjustable. The present invention is based on CMOS technology, the circuit structure is simple, and the sub-threshold region MOS tube is mostly used in the circuit, so that the neuron circuit has ultra-low power consumption characteristics; the input current of the circuit is pA level, and the output pulse frequency is Hz level, which can be very good It simulates the typical working state of biological neurons, effectively realizes the functions of neurons, and can be used to realize large-scale spiking neural network systems.

Description

technical field [0001] The invention belongs to the technical field of integrated circuit design, and in particular relates to a method for realizing a pulse neural network neuron circuit based on a LIF (Leaky Integrate and Fire) model. Background technique [0002] Biological systems are very energy efficient, especially the brain, which consists of billions of neuron cells and consumes about 20W. These units are noisy, imprecise, unreliable analog devices. When integrated into brain structures composed of interacting neurons, they can solve complex tasks and exhibit complex behaviors in real time and with high precision at very low power consumption. There are 1011 neurons and 1015 synapses in the human brain and the energy consumption is extremely low. If you want to simulate the parallel processing of information by a large number of neurons and synapses in the brain, the circuit needs to have ultra-low power performance and occupy as small an area as possible. The di...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a method for realizing a pulse neural network neuron circuit based on the LIF model, so as to overcome the problem of poor working performance of the electronic system under low power supply voltage in the prior art, and on the basis of the typical LIF model, make the neuron circuit With a refractory period, adjust the refractory period by adjusting the bias voltage, thereby controlling the pulse transmission frequency of the neuron circuit, so that the neuron circuit can adapt to working conditions at various frequencies

Images