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

Multi-label federated learning method based on tree structure, controller and medium

A tree structure, multi-label technology, applied in the computer field, can solve the problems of low model accuracy and neglect of mutual relations, etc., to achieve the effect of improving model accuracy and execution speed

Active Publication Date: 2021-07-23
上海嗨普智能信息科技股份有限公司 +1
View PDF8 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, this treatment actually ignores the relationship between the two labels (for example, the label "whether the loan will be repaid on schedule" may be able to help judge the label "whether the user will buy a product"), resulting in low model accuracy

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
  • Multi-label federated learning method based on tree structure, controller and medium
  • Multi-label federated learning method based on tree structure, controller and medium

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0020] This embodiment provides a multi-label federated learning method based on a tree structure, including:

[0021] Step S1. Obtain the training data set corresponding to each data holder among the n data holders. It can be understood that the respective training data sets of each data holder are stored locally. During the model training process, The respective training data sets of each data holder are always stored locally. The users of n data holders are overlapping, each data holder corresponds to a user label, and the i-th data holder corresponds to the first The i training data set is (X i ,Y i ),in, x i Denotes the sample user feature dataset of i training dataset, including n i sample users, each sample user has m i attribute; Y i Indicates the sample user label data set of the i training data set, i represents the serial number of the data holder, i=1,2,...,n, the y ik ∈R,k=1,2,...,n i ;

[0022] Wherein, it can be understood that the sample users of ...

Embodiment 2

[0066] The embodiment of the present invention also provides a multi-label-based federated learning data processing method, including:

[0067] Step C1. Obtain the training data set corresponding to each data holder among the n data holders. It can be understood that the respective training data sets of each data holder are stored locally. During the model training process, The respective training data sets of each data holder are always stored locally. The users of n data holders are overlapping, each data holder corresponds to a user label, and the i-th data holder corresponds to the first The i training data set is (X i ,Y i ),in, x i Denotes the sample user feature dataset of i training dataset, including n i sample users, each sample user has m i attribute; Y i Indicates the sample user label data set of the i training data set, i represents the serial number of the data holder, i=1,2,...,n, the y ik ∈R,k=1,2,...,n i ;

[0068] Wherein, it can be understood ...

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 relates to a multi-label federated learning method based on a tree structure, a controller and a medium. The method comprises the steps of S1, acquiring a training data set corresponding to each data holder in n data holders; S2, generating a label dependency relationship tree based on the user labels of the n data owners; S3, taking [X1, X2-Xn] union G (Yt (j)) as input data of a prediction sub-model Mt (j), taking the prediction label Yt (j) as output data of the prediction sub-model Mt (j), carrying out longitudinal federated learning training, and carrying out parallel training to generate the prediction sub-model Mt (j); S4, wherein the features of a to-be-tested user are [x1, x2-xn], all labels [y < t > (1), y < t > (2)- y < t > (n)] of the to-be-tested user are generated based on the [x1, x2- xn], the label dependency relation tree and the Mt (j), and y < t > (j) represents the predicted value of the to-be-tested user corresponding to the t (j) th label. According to the invention, federated learning is carried out based on the mutual relation among multiple labels, and the model precision and the model training speed are improved.

Description

technical field [0001] The invention relates to the field of computer technology, in particular to a tree-structure-based multi-label federated learning method, controller and medium. Background technique [0002] Vertical federated learning refers to federated learning in the case where users in the data sets of multiple data holders overlap more and user features overlap less. The task of vertical federated learning is to jointly train a machine learning model while maintaining data localization. Multi-label learning solves the technical problem of machine learning where one sample corresponds to multiple labels. For example, a picture may contain both dogs and cats. In contrast, traditional single-label learning solves the technical problem of machine learning where one sample corresponds to only one label, for example: judging whether a picture is a photo of a cat or a photo of a dog. [0003] Traditional single-label-based longitudinal federated learning has been rese...

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): G06K9/62G06N20/00
CPCG06N20/00G06F18/24323G06F18/214
Inventor 蔡文渊张坤坤高明周傲英徐林昊顾海林孙嘉
Owner 上海嗨普智能信息科技股份有限公司