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

Processor and training method for convolutional neural network

A convolutional neural network and training method technology, applied in neural learning methods, biological neural network models, neural architectures, etc., can solve the problems of reducing model size, increasing performance overhead, and not being able to effectively reduce memory usage

Active Publication Date: 2020-02-18
CAMBRICON TECH CO LTD
View PDF14 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Available memory size becomes a bottleneck limiting neural network models
[0004] Existing techniques usually reduce the size of the model, because the weight is not the main memory usage in neural network training, so the memory usage cannot be effectively reduced
Or copy the data structure back and forth between the memory of the central processing unit (Central Processing Unit, abbreviated as CPU) and the memory of the graphics processing unit (Graphics Processing Unit, abbreviated as GPU), which will increase performance overhead

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
  • Processor and training method for convolutional neural network
  • Processor and training method for convolutional neural network
  • Processor and training method for convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0086] In order to make the objectives, technical solutions, and advantages of the present disclosure clearer, the present disclosure will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0087] It should be noted that in the drawings or description of the specification, similar or identical parts use the same drawing numbers. The implementations not shown or described in the drawings are those known to those of ordinary skill in the art. In addition, although this article may provide an example of a parameter containing a specific value, it should be understood that the parameter does not need to be exactly equal to the corresponding value, but can be approximated to the corresponding value within an acceptable error tolerance or design constraint. In addition, the directional terms mentioned in the following embodiments, such as "up", "down", "front", "rear", "left", "right", etc., are only the dire...

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

The present disclosure provides a processor and a training method of a convolutional neural network; wherein, the processor of the convolutional neural network includes: an encoding module for encoding input data or output data of an activation layer; a calculation module, It is connected with the coding module, and is used to perform operations from forward propagation and back propagation; wherein, during back propagation, the calculation module is used to perform operations on partial derivatives according to coding results. The disclosed convolutional neural network processor and training method effectively save memory, reduce the number of input and output to the memory, optimize the performance of the convolutional neural network, and ensure the accuracy of the convolutional neural network prediction.

Description

[0001] The present invention is a divisional application of the invention patent application filed on May 18, 2018 with the application number 201810486460.8 and the invention title "Encoding Storage Device and Method, Processor and Training Method". Technical field [0002] The present disclosure relates to the field of artificial intelligence technology, and in particular to a processor and training method of a convolutional neural network. Background technique [0003] Convolutional Neural Network (Convolutional Neural Network, abbreviated as CNN) is a feed-forward neural network. Its artificial neurons can respond to a part of the surrounding units in the coverage area. In recent years, it has been widely used in image processing, speech processing, and pattern recognition And other fields. The availability of powerful data resources and its own good parallelism have made convolutional neural networks develop rapidly and have received widespread attention. As the number of ne...

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
Patent Type & Authority Patents(China)
IPC IPC(8): G06N3/04G06N3/063G06N3/08
CPCG06N3/084G06N3/063G06N3/045
Inventor 不公告发明人
Owner CAMBRICON TECH CO LTD