Data processing method and device and related equipment
A data processing and data technology, applied in the computer field, can solve problems such as reducing data processing efficiency, increasing data arrangement memory reading time, and failing to fully utilize the bandwidth and computing power of intelligent processing devices.
Pending Publication Date: 2022-01-11
HUAWEI TECH CO LTD
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, when the neural network model processes multiple channels of data, it needs to perform operations such as vectorization and fractalization on each channel of data, which will increase the time for data arrangement (tiling) and memory reading, and reduce the efficiency of data processing
And if the amount of data in each channel is small, compared to processing data with a larger amount of data in one forward reasoning process, processing multiple channels of data with a smaller amount of data in one forward reasoning cannot make full use of the bandwidth and bandwidth of the intelligent processing device. Calculate ability
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0041] The control system and method provided by the present application will be described in detail below with reference to the accompanying drawings.
[0042] A convolutional neural network is a deep learning model that is often used to analyze data such as images. Such as figure 1 As shown, CNN usually includes a convolutional neural network including convolutional layers, pooling layers, and fully connected layers. The convolutional layer is used to aggregate the features of the input data to obtain the feature map; the pooling layer is used to select the feature map output by the convolutional layer after the feature extraction of the convolutional layer; The features obtained by the pooling layer and the pooling layer are combined nonlinearly to obtain the output result. Each network layer includes network layer parameters such as kernel size, padding size, and stride size. For example, the convolution parameters corresponding to the convolutional layer include the co...
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
Login to View More Abstract
The invention discloses a data processing method. The method comprises the steps of enabling the intelligent processing device to obtain first data and second data; filling third data used for isolating the first data from the second data between the first data and the second data according to a preset rule to obtain fourth data; and processing the fourth data by using the convolutional neural network, wherein the first data and the second data are data to be spliced together, the sequence of the first data is earlier than that of the second data during splicing, and the first data and the second data are any one of image data, voice data or text sequences. The two groups of data are spliced into the data which comprises more elements and is relatively large in data volume, and the data obtained after splicing is input into the convolutional neural network model for processing, so that the efficiency of processing the data by the intelligent processing device by using the convolutional neural network can be improved, and the bandwidth of the intelligent processing device can be fully utilized.
Description
technical field [0001] The present application relates to the field of computer technology, in particular to a data processing method, device and related equipment. Background technique [0002] In the fields of image processing, natural language processing, etc., it is often necessary to process multiple channels of data at the same time. When using convolutional neural networks (CNN) to process multiple channels of data, in order to improve the efficiency of data processing by intelligent processing devices, usually The multi-channel data is formed into a group (batch) and input into the convolutional neural network together to realize a forward reasoning to complete the processing of the multi-channel data. However, when the neural network model processes multiple channels of data, it needs to perform vectorization and fractal operations on each channel of data, which will increase the time for data tiling and memory reading, and reduce the efficiency of data processing. ...
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
Login to View More IPC IPC(8): G06K9/62G06N3/04G06T3/40
CPCG06T3/4038G06T2200/32G06N3/045G06F18/251G06F18/25G06V10/803G06V10/82G06N3/0464G06V10/77G06N3/091
Inventor 王永忠魏富春张伟徐晓忻
Owner HUAWEI TECH CO LTD