(Qu ality + g nomon ) An open testbed that contains a toolset to support users for the quality management of AI systems .
Download (Free)
(Qu ality + g nomon ) An open testbed that contains a toolset to support users for the quality management of AI systems .
Download (Free)Qunomon provides a holistic environment to test, evaluate, and report the quality of AI systems based on machine learning components. The Machine Learning Quality Management Guideline states a technical basis for quality goals for machine learning-based products/services. Qunomon shows the best practice to implement/evaluate/explain the quality of machine learning systems harmonizing with the guideline.
The Qunomon is equipped with a Web application that automatically generates the quality reports from the testing results with a machine learning model and datasets. The application user can easily test various testing programs, set acceptance criteria, and evaluate the quality of models and datasets according to the guideline. The generating report will help users to encourage communication between quality evaluators and AI system developers to improve the quality of ML components that are essential.
The Qunomon manages various types of testing methods with the concept of AIT. An AIT is a unit to abstract a software program that measures specific quality metrics, sample datasets, and machine learning algorithms for the development of AI systems. The Qunomon provides common APIs of deployed AIT on the system to access the testing results from a user application. AIT_SDK allows test method developers to build and deploy a new AIT.
Qunomon is an open, adaptable, and flexible framework for the quality evaluation of ML components by employing new testing methods for machine learning algorithms depending on the business requirements.