Learning data generation apparatus
The learning data generation device addresses inefficiencies in training data preparation by adding noise and using a classifier to enhance the quality of training data, thereby improving the efficiency of machine learning models.
JP2025175428AActive Publication Date: 2025-12-03奥野 修二
Patent Information
- Application Number
- JP2024081533
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-05-20
- Publication Date
- 2025-12-03
- Estimated Expiration
- 2044-05-20
AI Technical Summary
Technical Problem
The quality and quantity of training data significantly impact the performance of machine learning models, with preparing high-quality datasets being labor-intensive and the presence of unsuitable data leading to inefficiencies.
Method used
A learning data generation device that includes an input unit for reference image data, a degradation processing unit to add noise, a generator to restore degraded data, and a classifier to distinguish between true and false data, ensuring higher-quality training data is prepared.
Benefits of technology
The device enhances learning efficiency by uniformly preparing higher-quality training data, improving the performance of machine learning models.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure 2025175428000001_ABST
Abstract
To provide a learning data generation apparatus configured to uniformly prepare high-quality teacher data so as to improve learning efficiency of a machine learning model in data processing using machine learning.SOLUTION: A learning data generation apparatus includes: an input unit 110 which receives an input of reference image data created for reducing noise components; a degradation processing unit 111 which creates degraded reference image data by adding noise to the reference image data input to the input unit 110; a generator 112a which learns setting values in a machine learning model 112 to be learned so that the degraded reference image data can be restored to the reference image data, using, as input, the degraded reference image data generated in the degradation processing unit 111; and a learning processing execution unit 101 which executes the processing in the generator 112a. An inference apparatus outputs image data obtained by the machine learning model 112 and input to the inference apparatus, as teacher image data for another machine learning.SELECTED DRAWING: Figure 2
Need to check novelty before this filing date? Find Prior Art
Citation Information
Patent Citations
Image processing apparatus and program
JP2020154605A
Diagnosis support device, method for processing image on diagnosis support device and program
JP2018038789A
Super-resolution device and program therefor
JP2020027557A