Neuron coding circuit

Abstract

The invention relates to a neuron coding circuit, belonging to the technical field of integrated circuits. In the neuron coding circuit, the analog voltage signal output by the traditional neural network is filtered and processed by a signal acquisition module to output a stable control voltage signal; The control voltage signal and the current signal outputted from the sensitivity adjusting unitare input to a frequency encoder, and after quantized and encoded by the frequency encoder, an analog signal with sigmoid function relationship between the frequency and the control voltage signal isoutputted. The pulse width regulator receives the analog signal and the pulse width control signal output from the frequency encoder, and outputs the pulse signal whose frequency is the same as the analog signal and whose pulse width is positively related to the pulse width control signal. The pulse emitting module receives the pulse signal from the pulse width regulator and transmits the pulse signal to the pulse neural network when the signal Ready emitted from the membrane potential detecting module is enabled. The neuron coding circuit of the invention has the advantages of simple circuitstructure, wide frequency coverage range, low power consumption and the like.

Description

technical field [0001] The invention belongs to the technical field of integrated circuits, and in particular relates to a neuron encoding circuit. Background technique [0002] A neural network is a mathematical model or computational model that imitates the structure and function of a biological neural network, and is mainly calculated based on a large number of artificial neuron connections. Usually, the artificial neural network can change the internal structure on the basis of external information, which is an adaptive system. [0003] The traditional neural network is an operation model, which is composed of a large number of nodes connected to each other. Each node represents a specific output function, called the activation function; every connection between two nodes represents a weighted value for the signal passing through the connection, called the weight, which is equivalent to the memory of the artificial neural network. The output of the network is different...

AI Technical Summary

Problems solved by technology

[0006] The object of the present invention is to propose a neuron coding circuit for the problem of difficult training of the spike neural network mentioned in the background technology, which forms a pulse neural network identifiable neural network by encoding the value of a neuron in the traditional neural network. Pulse signal, simple circuit structure, low power consumption

Images