Machine learning programs, methods, and apparatus

By calculating independence using mutual information to select data for labeling, the method addresses the inefficiencies of conventional fair active learning, optimizing fairness and accuracy trade-offs, and reducing processing loads in machine learning model training.

JP7882356B2Active Publication Date: 2026-06-30FUJITSU LTD
View PDF 5 Cites 0 Cited by

Patent Information

Authority / Receiving Office
JP · JP
Patent Type
Patents
Current Assignee / Owner
FUJITSU LTD
Filing Date
2023-02-09
Publication Date
2026-06-30

AI Technical Summary

Technical Problem

Conventional fair active learning methods for training machine learning models are computationally intensive and inefficient, particularly for complex models, leading to high processing loads due to the need for extensive data selection and training processes.

Method used

A method that calculates the independence between prediction results and protected attributes using mutual information to select data for labeling, reducing the need for retraining by evaluating fairness without extensive model training, and optimizing the trade-off between fairness and accuracy.

Benefits of technology

Reduces processing load and execution time while maintaining fairness and accuracy improvements in machine learning models, enabling efficient data selection and training.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 0007882356000005
    Figure 0007882356000005
  • Figure 0007882356000006
    Figure 0007882356000006
  • Figure 0007882356000007
    Figure 0007882356000007
Patent Text Reader

Abstract

Provided is a machine learning device that: calculates the independence between a prediction result, which is obtained when inputting each of a plurality of pieces of unlabeled data into a machine learning model, and the value of a first attribute of each of the plurality of pieces of data; selects a first piece of data from the plurality of pieces of data on the basis of the independence; obtains a label for the first piece of data; and executes training of the machine learning model on the basis of the first piece of data and the label.
Need to check novelty before this filing date? Find Prior Art