Cyber Physical Security Research Center (Director: MATSUMOTO Tsutomu) and Artificial Intelligence Research Center (Director: TSUJII Junichi) of National Institute of Advanced Industrial Science and Technology (AIST) jointly published first English version of “Machine Learning Quality Management Guideline”.
This document is a translation of the corresponding Japanese version, 1st edition, published on June 30, 2020 (press release in Japanese).

The guideline document is available from the website of Cyber Physical Security Research Center :
https://www.cpsec.aist.go.jp/achievements/aiqm/index_en.html

This document establishes a basis for quality goals for machine learning-based products/services, and provides procedural guidance for realizing quality through development process management and system evaluations.  Quality goals are set for three categories (safety, performance, and fairness) and classified by the nature of target products/services.  Quality management for each of their goals is eight management/evaluation criteria stated in the document.

The guideline is a fruit of discussions by “Committee for machine learning quality management” organized of industrial-sector engineers and academic researchers.  The project is supported by NEDO.