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

A model parameter training method and device

A technology of model parameters and training methods, applied in the field of machine learning, can solve the problems of not being able to give, relying on a large number of labeled samples, and scarce labeled samples, and achieve the effect of saving labor costs

Active Publication Date: 2022-05-10
北京如布科技有限公司
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, some machine learning models have no upper limit on the prediction value of new data, so there is no reasonable threshold to filter "pseudo-labeled samples", so in this case it is still impossible to use unlabeled samples to alleviate the first A shortcoming of the scarcity of labeled samples in the prior art
[0007] There are two shortcomings in the existing technology: either rely on a large number of labeled samples, or need to provide a threshold to use unlabeled samples to alleviate the problem of scarce labeled samples

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 model parameter training method and device
  • A model parameter training method and device
  • A model parameter training method and device

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0017] figure 1 It is a flow chart of the model parameter training method provided by Embodiment 1 of the present invention. This embodiment is applicable to machine learning. The method can be executed by a model parameter training device, which can be implemented by software and / or hardware. The device can Integrate in any device that provides artificial intelligence, such as typical user terminal devices, such as mobile phones, tablets, smart TVs or smart watches. The method includes:

[0018] S101. Acquire a sample data set and a data model for prediction according to the sample data set, the sample data set includes labeled samples and unlabeled samples, and the data model includes an initialized vector of the first parameter V and a second parameter A vector of U.

[0019] Further, the labeled sample is (x, y), the unlabeled sample is z, and the first parameter V is a vector w v , the second parameter U is a vector w u .

[0020] Suppose there is a set of labeled sa...

Embodiment 2

[0032] figure 2 It is a flow chart of the model parameter training method provided by Embodiment 2 of the present invention, and Embodiment 2 is based on Embodiment 1. The model parameter training method provided in this embodiment includes the following steps: step S201, step S202 and step S203. Wherein, step S202 is the optimization of step S102 in the first embodiment, step S203 is the optimization of the step S103 in the first embodiment, step S201 is the same as step S101 in the first embodiment, and the same steps will not be repeated.

[0033] S201. Obtain a sample data set and a data model for prediction based on the sample data set, the sample data set includes labeled samples and unlabeled samples, and the data model includes an initialized vector of the first parameter V and a second parameter A vector of U.

[0034] S202. For the marked sample set D L Each labeled sample in : by the first feature vector Φ of the labeled sample v and the first parameter V to ca...

Embodiment 3

[0047] image 3 It is a schematic structural diagram of the model parameter training device provided by Embodiment 3 of the present invention. The device is used to execute the method for training model parameters in the above embodiments. The device includes: an acquisition module 301 , a first training module 302 and a second training module 303 .

[0048] An acquisition module 301, configured to acquire a sample data set and a data model for prediction according to the sample data set, the sample data set includes labeled samples and unlabeled samples, and the data model includes a vector of initialized first parameters V and a vector of the second parameter U.

[0049] The first training module 302 is connected to the acquisition module 301 and configured to train the first parameter V and the second parameter U according to the labeled samples.

[0050] The second training module 303 is connected with the first training module 302, and is used for retraining the first ...

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 embodiment of the invention discloses a model parameter training method and device. The method includes: acquiring a sample data set and a data model for prediction according to the sample data set, the sample data set includes labeled samples and unlabeled samples, and the data model includes an initialized vector of the first parameter V and the first parameter V Two vectors of parameters U; training the first parameter V and the second parameter U according to the labeled sample; training the first parameter V and the second parameter U according to the labeled sample, according to the labeled sample. The above unlabeled samples are used for training again. The technical solution provided by the embodiments of the present invention fully utilizes a large number of unlabeled samples and a small number of labeled samples to train model parameters for a prediction model that cannot provide an upper limit of the predicted value, saving a lot of labor costs for labeling samples.

Description

technical field [0001] Embodiments of the present invention relate to machine learning technology, and in particular to a method and device for training model parameters. Background technique [0002] A model that predicts output based on input and parameters can be used in artificial intelligence, for example, to accurately identify the file that the user needs based on the colloquial text input by the user, or to accurately identify the object in the picture based on the input image. For general machine learning algorithms, the performance of the model mainly depends on its parameter configuration. Models generated with different parameter combinations often have large performance differences. Usually the parameters of the model need to be trained. [0003] The basic definition of parameter training is as follows: Given a training data set X T , the goal of parameter training is to find a parameter combination θ of a machine learning algorithm F, in X T Build a model f...

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): G06N20/00G06K9/62
CPCG06N20/00
Inventor 吉宗诚王君保郭祥郭瑞雷宇
Owner 北京如布科技有限公司