Empowering the trustworthiness of ML-based critical systems through engineering activities

TL;DR

This paper reviews the engineering process for developing trustworthy ML algorithms in critical systems, emphasizing design principles, data handling, implementation, evaluation, and deployment within a comprehensive framework.

Contribution

It introduces a unified framework for designing trusted ML systems, integrating core engineering activities from domain specification to deployment.

Findings

Framework consolidates trust-related engineering activities

Highlights importance of design and evaluation in trustworthiness

Provides guidelines for deploying reliable ML in critical systems

Abstract

This paper reviews the entire engineering process of trustworthy Machine Learning (ML) algorithms designed to equip critical systems with advanced analytics and decision functions. We start from the fundamental principles of ML and describe the core elements conditioning its trust, particularly through its design: namely domain specification, data engineering, design of the ML algorithms, their implementation, evaluation and deployment. The latter components are organized in an unique framework for the design of trusted ML systems.