Information technology — Artificial intelligence — Treatment of unwanted bias in classification and regression machine learning tasks

Scope

This document provides mitigation techniques that can be applied throughout the AI system life cycle in order to treat unwanted bias. This document describes how to address unwanted bias in AI systems that use machine learning to conduct classification and regression tasks. This document is applicable to all types and sizes of organization. ©ISO/IEC 2022. All rights reserved.

Purpose

No international standard currently defines the appropriate steps that should be taken to treat unwanted bias during the development, or use, of AI systems.

This proposal is based on ISO/IEC TR 24027 and provides mitigation techniques in accordance with the life cycle as defined in ISO/IEC 22989 and ISO/IEC 5338. The mitigation techniques in the document will be agnostic of context, for example fairness. The document also formalizes the taxonomy of bias described in ISO/IEC TR 24027.

The proposal is scoped to include the treatment of unwanted bias in ML classification and regression systems as most of the expert contributions submitted during the development of ISO/IEC TR 24027 have been about these systems. See, for example, Clauses 5 and 7 of ISO/IEC TR 24027.

This document provides good practice on treating unwanted bias and can help industry comply with various regional and national regulatory requirements about the treatment of unwanted bias in the ML systems in scope. ©ISO/IEC 2022. All rights reserved.

External Links