Processor integrating artificial neural network and spiking neural network

Abstract

The invention discloses a processor integrating an artificial neural network and a pulse neural network, and belongs to the field of artificial intelligence hardware. The system comprises a shared storage unit, a main controller, an ANN learning control circuit, an SNN reasoning control circuit and a shared calculation unit. And the weight output by the ANN learning control circuit is used as the weight of the SNN reasoning control circuit to realize the fusion of the two neural networks, so that the processor has two working modes in a high-accuracy scene and a low-power-consumption environment. During use, the ANN learning control circuit or the SNN reasoning control circuit is controlled by the main controller to conduct reasoning according to actual scene requirements, the problem that existing signal classification processing based on a neural network cannot be suitable for complex and changeable processing scenes is solved, and meanwhile the accuracy of input monitoring in a low-power-consumption environment is improved.

Description

technical field [0001] The invention relates to the field of artificial intelligence hardware, in particular to a processor integrating artificial neural network and impulse neural network. Background technique [0002] Thanks to the improvement of computer computing power, Artificial Neural Networks (ANNs), especially Convolutional Neural Networks, perform well in various classification tasks. The classification algorithm of ANN based on gradient descent training has become the current mainstream detection method because of its high accuracy. However, affected by the characteristics of its algorithm itself, ANN needs to update all the neuron states of the network every time it is updated. Problems with event-driven characteristics, such as ECG detection, keyword detection, EEG signal detection, etc., lead to a lot of waste of computing resources. Because in continuous and uninterrupted signal waveform detection, abnormal signals are very sparse, most of them are normal sig...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a processor that integrates artificial neural network and pulse neural network, so as to solve the problem that the existing neural network cannot be applied to complex and changeable processing scenarios and the differences and low energy consumption requirements of different personnel.

Images