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

A method and apparatus for filling matrix boundaries in a convolution neural network

A technology of convolutional neural network and filling method, which is applied in the field of filling and device of matrix boundary in convolutional neural network, which can solve the problem of low convolution precision and achieve the effect of improving precision and expanding size

Pending Publication Date: 2019-02-01
旺微科技(上海)有限公司
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, in the original matrix, each element is very important. After many elements are discarded, the elements participating in the convolution will be reduced, thereby reducing the amount of information in the convolution process, resulting in low convolution accuracy. technical issues

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
  • A method and apparatus for filling matrix boundaries in a convolution neural network
  • A method and apparatus for filling matrix boundaries in a convolution neural network
  • A method and apparatus for filling matrix boundaries in a convolution neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0041] The embodiments of the present invention are described in detail below. This embodiment is implemented on the premise of the technical solution of the present invention, and detailed implementation methods and specific operating procedures are provided, but the protection scope of the present invention is not limited to the following implementation example.

[0042] Embodiments of the present invention provide a method and device for filling a matrix boundary in a convolutional neural network. The following firstly introduces a method for filling a matrix boundary in a convolutional neural network provided by an embodiment of the present invention.

[0043] figure 1 It is a schematic flowchart of a method for filling matrix boundaries in a convolutional neural network provided by an embodiment of the present invention, figure 2 A schematic diagram of the principle of matrix convolution in a convolutional neural network provided by an embodiment of the present inventio...

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 invention discloses a filling method of matrix boundary in convolution neural network, The method comprises the following steps: 1) according to the position of the elements to be calculated in the original matrix before convolution and the size of the kernel matrix of the convolution neural network, judging whether there are elements that cannot be convolution in the convolution process, wherein, the original matrix comprises a convolution matrix or a pooling matrix; 2), if yes, adding preset elements of first preset number column and / or second preset number row outside the boundary of original matrix before convolution. The invention discloses a filling device of matrix boundary in convolution neural network.

Description

technical field [0001] The present invention relates to a method and a device for processing a boundary of a convolutional neural network, and more particularly to a method and a device for filling a matrix boundary in a convolutional neural network. Background technique [0002] CNN (Convolutional Neural Network, Convolutional Neural Network) is a feedforward neural network. At present, CNN has become one of the research hotspots in many scientific fields, especially in the field of pattern classification. Because the network avoids the complex preprocessing of images and can directly input original images, it has been more widely used. Since the feature detection layer of CNN learns through training data, when using CNN, it avoids explicit feature extraction and learns implicitly from training data; moreover, because the neuron weights on the same feature map Same, so the network can learn in parallel, which is also a big advantage of the convolutional network over the 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 Applications(China)
IPC IPC(8): G06F17/16
CPCG06F17/16
Inventor 曾建徐昕肖立波
Owner 旺微科技(上海)有限公司