Model training method and device, task processing method, storage medium and processor

A technology for model training and task processing, applied in the field of deep learning, can solve problems such as poor anti-risk active perception and resilience, and inability to achieve model training, so as to improve active perception and resilience, deal with model performance degradation, and improve robustness. self-enhancing effect

Pending Publication Date: 2022-03-11
ALIBABA (CHINA) CO LTD
View PDF0 Cites 3 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Embodiments of the present invention provide a model training method and device, a task processing method, a storage medium, and a processor, to at least solve the problem that th

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
  • Model training method and device, task processing method, storage medium and processor
  • Model training method and device, task processing method, storage medium and processor
  • Model training method and device, task processing method, storage medium and processor

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0029] According to an embodiment of the present invention, an embodiment of a model training method is provided. It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and, although in The flowcharts show a logical order, but in some cases the steps shown or described may be performed in an order different from that shown or described herein.

[0030] The method embodiment provided in Embodiment 1 of the present invention may be executed in a mobile terminal, a computer terminal, or a similar computing device. figure 1 A hardware structural block diagram of a computer terminal (or mobile device) for implementing the model training method is shown. Such as figure 1 As shown, the computer terminal 10 (or mobile device 10) may include one or more (shown by 102a, 102b, ..., 102n in the figure) processor 102 (the processor 102 may include but not limited to a mi...

Embodiment 2

[0087] According to an embodiment of the present invention, an embodiment of a device for implementing the above model training method is also provided, Figure 5 According to a schematic structural diagram of a model training device according to an embodiment of the present invention, as shown in 5, the above device includes: a first training module 500, an anti-disturbance module 502 and a second training module 504, wherein:

[0088] The first training module 500 is used to obtain the first model by using the first data training, wherein the first data is labeled data, and the first model is a teacher model; the anti-disturbance module 502 is used for The above-mentioned first model performs anti-disturbance to obtain second data, wherein the above-mentioned second data is pseudo-labeled data; the second training module 504 is used to train the above-mentioned first data and the above-mentioned second data to obtain the second model, wherein , the above second model is the ...

Embodiment 3

[0092]According to an embodiment of the present invention, an embodiment of an electronic device is also provided, and the electronic device may be any computing device in a computing device group. The electronic device includes: a processor and a memory, wherein:

[0093] A processor; and a memory, connected to the above-mentioned processor, for providing the above-mentioned processor with instructions for processing the following processing steps: Step 1, using the first data training to obtain the first model, wherein the above-mentioned first data is labeled data, The above-mentioned first model is a teacher model; step 2, based on the above-mentioned first data, the above-mentioned first model is subjected to anti-disturbance to obtain second data, wherein the above-mentioned second data is pseudo-labeled data; step 3, using the above-mentioned first data and the above-mentioned second data training to obtain a second model, wherein the above-mentioned second model is a s...

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 model training method and device, a task processing method, a storage medium and a processor. The method comprises the steps that first data is used for training to obtain a first model, the first data are data with labels, and the first model is a teacher model; performing anti-disturbance on the first model based on the first data to obtain second data, the second data being data with pseudo labels; and training by adopting the first data and the second data to obtain a second model, the second model being a student model. The method solves the technical problems that a model training method in the related technology cannot realize model training under a small amount of marked data, and the active perception and defending capability of the anti-risk is poor.

Description

technical field [0001] The present invention relates to the technical field of deep learning, in particular to a model training method and device, a task processing method, a storage medium and a processor. Background technique [0002] As deep learning technology continues to make new breakthroughs, artificial intelligence models based on deep neural networks are widely used in various fields such as images, text, and speech. Existing artificial intelligence models usually need to rely on a large number of high-quality, labeled The data. [0003] On the one hand, it is often time-consuming and labor-intensive to obtain high-quality labeled data in real-world scenarios, requiring a lot of labor and material costs, and the quality of labels will also be greatly affected by the learning ability, prior knowledge and The influence of the difficulty of the labeling task itself. On the other hand, existing artificial intelligence models are often built based on clean, non-advers...

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): G06K9/62
CPCG06F18/2155G06F18/2415
Inventor 李进锋
Owner ALIBABA (CHINA) CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap