Title:
AI QUALITY MONITORING SYSTEM
Document Type and Number:
WIPO Patent Application WO/2022/259592
Kind Code:
A1
Abstract:
The objective of the present invention is to provide a technique enabling monitoring of the quality of a learning model without depending on input data and efficiently. An AI quality monitoring system according to the present invention acquires an inferred result obtained by inference performed by a comparative model on input data of the same type as learning data learned by a learning model, and compares the inferred result acquired from the comparative model with correct answer data to calculate an evaluation value indicating a degree of suitability of the comparative model for determination of the quality of the learning model (see FIG. 1).
Inventors:
ITO HIDEKAZU (JP)
Application Number:
PCT/JP2022/001352
Publication Date:
December 15, 2022
Filing Date:
January 17, 2022
Export Citation:
Assignee:
HITACHI SOLUTIONS LTD (JP)
International Classes:
G06N20/00
Domestic Patent References:
WO2021079485A1 | 2021-04-29 |
Foreign References:
JP2009140283A | 2009-06-25 | |||
US10810512B1 | 2020-10-20 |
Attorney, Agent or Firm:
HIRAKI & ASSOCIATES (JP)
Download PDF: