Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Deep learning network training method based on artificial intelligence

A deep learning network and artificial intelligence technology, applied in the fields of equipment and storage media, systems, and deep learning network training methods, can solve the problems of fitting, neural network models are difficult to guarantee results, parameter redundancy, etc.

Pending Publication Date: 2021-11-02
ZHENGZHOU UNIV
View PDF4 Cites 5 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, traditional deep learning models usually contain a large number of redundant parameters, which are prone to overfitting problems, making it difficult for neural network models to guarantee optimal results during training.
At present, there is still a lack of a good training mechanism in the prior art to solve the above problems

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
  • Deep learning network training method based on artificial intelligence
  • Deep learning network training method based on artificial intelligence
  • Deep learning network training method based on artificial intelligence

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0026] see figure 1 A kind of deep learning network training method based on artificial intelligence that the embodiment of the present application provides, comprises the following steps:

[0027] S101. Construct a neural network;

[0028] The steps of constructing the neural network include establishing the input layer, hidden layer and output layer of the neural network; determining the activation function of the neural network.

[0029] Artificial neural network is a parallel distributed system, which adopts a completely different mechanism from traditional artificial intelligence and information processing technology, overcomes the defects of traditional logical symbol-based artificial intelligence in processing intuition and unstructured information, and has self-adaptive, Features of self-organization and real-time learning.

[0030]The concept of deep learning comes from artificial neural networks, and the multi-layer perceptron with multiple hidden layers is a deep ...

Embodiment 2

[0069] see figure 2 , in some embodiments of the present invention, the above-mentioned training set includes question items and answer items corresponding to the above-mentioned question items; in step S103, the process of training the above-mentioned neural network on the above-mentioned training set is:

[0070] S201. Input the data of the above-mentioned training set into the above-mentioned neural network, and the above-mentioned neural network obtains an output value through forward propagation calculation;

[0071] Forward propagation is the process of continuously calculating the weights and biases of each layer from the input layer through layers of hidden layers, and finally obtaining the output value y^.

[0072] S202. Calculate and obtain a loss function according to the above output value and the above answer item;

[0073] The loss is then calculated based on the difference between the output value y^ and the answer item y (true value). For the loss function, ...

Embodiment 3

[0080] In some embodiments of the present invention, in step S105, before adjusting the parameters of the selected sub-neural network model according to the verification result, it also includes: judging whether the sub-neural network model is overfitting or not according to the verification result Underfitting.

[0081] Before adjusting the parameters of the sub-neural network model, it is necessary to judge whether the model is under-fitting or over-fitting to facilitate subsequent optimization of the sub-neural network model.

[0082] Exemplarily, the easiest way to judge the current model is to compare its error on the training set with the error on the cross-validation set. When the cross-validation error is not much different from the training set error, and the training set error is large, the current model is more likely to be underfit; and when the cross-validation error is much larger than the training set error, and the training set error is small, the current model...

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 deep learning network training method, system and device based on artificial intelligence, and a storage medium, and relates to the technical field of artificial intelligence, the deep learning network training method based on artificial intelligence comprises the following steps: constructing a neural network; initializing parameters of the neural network; obtaining a verification set and a plurality of training sets, and training the neural network on the plurality of training sets to obtain a plurality of sub-neural network models; respectively verifying the plurality of sub-neural network models on the verification set; according to the accuracy of the verification result, selecting sub-neural network models with a fixed proportion from high to low, adjusting parameters of the selected sub-neural network models, and then enabling the selected sub-neural network models to continue training; and after training is completed, selecting an optimal sub-neural network model. According to the method and the system provided by the invention, the overfitting problem of the model is effectively solved, and the training performance and the accuracy of the model are improved.

Description

technical field [0001] The present invention relates to the technical field of artificial intelligence, in particular to an artificial intelligence-based deep learning network training method, system, device and storage medium. Background technique [0002] Deep learning solves many challenging problems, and its results have been widely used in computer vision, speech recognition, natural language processing and other fields. Technologies such as image recognition, video processing, and speech recognition based on deep learning have great application prospects and demands on end devices of edge computing systems. However, traditional deep learning models usually contain a large number of redundant parameters, which are prone to over-fitting problems, making it difficult for neural network models to guarantee optimal results during training. At present, there is still a lack of a good training mechanism in the prior art to solve the above problems. Contents of the inventio...

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): G06N3/04G06N3/08
CPCG06N3/08G06N3/084G06N3/045
Inventor 王俊凯
Owner ZHENGZHOU UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com