Convolutional neural network processing method and device

Abstract

The invention discloses a convolutional neural network processing method and device, and relates to the technical field of computers. The specific implementation scheme is as follows: obtaining an input matrix, a weight matrix, an activation identifier, an activation function and a weight matrix group number; when the group number of the weight matrixes is smaller than or equal to the number of the accumulators, respectively inputting the input matrix and the weight matrixes into a plurality of accelerator registers; controlling the plurality of arithmetic units to read the input matrix and the weight matrix from the plurality of accelerator registers, and generating a plurality of arithmetic results; and inputting the plurality of operation results into a plurality of accumulators for accumulation to generate an accumulation result, and activating the accumulation result according to the activation identifier and the activation function. According to the method, accumulation results generated by a plurality of accumulators are activated and are processed in parallel with the convolution calculation process, so that the calculation efficiency of the convolutional neural network isimproved, and the delay problem caused by the convolution calculation efficiency is remarkably improved.

Description

technical field [0001] The present application relates to the field of convolutional neural network technology in the field of computer technology, and in particular to a method and device for processing a convolutional neural network. Background technique [0002] Convolutional neural network is the key technology of deep learning. However, due to the characteristics of large order of magnitude, complex layers, and large depth of convolutional neural network model, there is a problem of low efficiency when using traditional processor units for convolution calculations; especially in speech For scenarios with sensitive delay requirements such as signal processing, the convolution calculation delay makes the real-time requirements of the system face a huge challenge. [0003] In related technologies, when the convolutional neural network is processed based on the specific parameters of the model, the activation process is performed intensively after the dot product calculatio...

AI Technical Summary

Problems solved by technology

Due to the slow speed of activation processing, the existing convolutional neural network has technical problems of low computational efficiency

Images