Unlock instant, AI-driven research and patent intelligence for your innovation.

Convolution neural network forward acceleration method, device and system

A convolutional neural network, convolutional neural technology, applied in the field of device and system, convolutional neural network forward acceleration method

Active Publication Date: 2019-01-01
TENCENT TECH (SHENZHEN) CO LTD
View PDF11 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] However, with the increasing use of artificial intelligence, the response speed of convolutional neural networks has gradually become a bottleneck restricting its application.

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Convolution neural network forward acceleration method, device and system
  • Convolution neural network forward acceleration method, device and system
  • Convolution neural network forward acceleration method, device and system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0025] In order to make the purpose, technical solutions and advantages of the present invention clearer, the following will further describe in detail the embodiments of the present invention in conjunction with the accompanying drawings.

[0026] The GoogLeNet architecture was developed by Christian Szegedy et al. From Google Research, won the ILSVRC 2014 challenge by getting an error rate 7% lower than the top 5. This great performance comes largely from the Inception module, which has a deeper network than previous convolutional neural networks (CNNs). This is achieved through a sub-network called the Inception Module, which allows GoogLeNet to use parameters more efficiently than previous architectures. In terms of actual parameter values, GoogLeNet has 10 times fewer parameters than AlexNet (about 6 million instead of 60 million ).

[0027] figure 1 Describes the Inception module in the GoogleNet neural network convolutional architecture. The notation "3×3+2(S)" mean...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

A method for accelerating the forward direction of convolution neural network includes such steps as splicing four sequential forward layers of convolution neural network: a convolution layer, a batchstandardization layer, a zoom layer and a non-linear activation layer to form a serial convolution layer, converging the forward layers to form a serial convolution layer, converging the forward layers to form a non-linear activation layer, converging the forward layers to form a series convolution layer, converging the forward layers to form a series convolution layer, converging the forward layers to form a series convolution layer. And / or the convolution cores of the same layer and volume in the convolution layer are spliced, such that the convolution cores of the same layer and volume arespliced into a parallel convolution core; and / or a feature map having a double dimensional relationship is extracted, the extracted feature maps are grouped, and the feature maps in the grouping areserially spliced.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to a convolutional neural network forward acceleration method, device and system. Background technique [0002] With the development of artificial intelligence (AI, Artificial Intelligence), various neural network convolution models have appeared in the prior art, including: LeNet-5 model, AlexNet model, VGGNet model, GoogleNet model, ResNet model and so on. Each neural network model has its characteristics, such as: [0003] The LeNet-5 model is the first convolutional neural network model successfully applied to digital recognition, and the convolutional layer has its own activation function. [0004] The AlexNet model is a more classic convolutional neural network model. Its structure is usually: input layer → (convolutional layer → pooling layer) → fully connected layer → output layer. The convolution kernel side length of the AlexNet convolutional layer...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
IPC IPC(8): G06N3/04G06N3/08
CPCG06N3/08G06N3/045
Inventor 陈志博李彦融黄小明李集佳石楷弘
Owner TENCENT TECH (SHENZHEN) CO LTD