Resources

Everything You Need to Implement ML FMEA

Access research papers, templates, tools, and implementation guides to help you apply the ML FMEA methodology in your organization.

Related Standards & Guidelines

ML FMEA aligns with and supports compliance with these industry standards.

ISO 21448 (SOTIF)

Safety of the Intended Functionality

Addresses functional insufficiencies in ML specification and measurement data. ML FMEA helps satisfy the requirement for "analysis of off-line training process of machine learning algorithms."

UL 4600

Standards for Safety for the Evaluation of Autonomous Products

Covers V&V procedures, engineering rigor, and data quality requirements for ML algorithms in autonomous vehicles. ML FMEA provides evidence for suitable engineering rigor.

ISO PAS 8800

Road Vehicles – Safety and Artificial Intelligence

Prescribes analyses to ensure engineering rigor in ML pipeline development. ML FMEA methodology directly supports these requirements.

ISO TR 5469

Artificial Intelligence — Functional Safety and AI Systems

Lists fault model methodologies including process-level FMEA. ML FMEA provides the specific implementation guidance for this requirement.

Implementation Tools & Templates

Practical tools to help you implement ML FMEA in your organization.

ML FMEA Template

Pre-populated template with ML Pipeline steps, potential failure modes, causes, and current best practices. Ready to use for your specific ML project.

Download Template

Implementation Checklist

Step-by-step checklist to guide your team through the ML FMEA process, ensuring no critical steps are missed.

Coming Soon

Stakeholder Guide

Guidance on identifying and engaging the right stakeholders for effective ML FMEA implementation across different roles and departments.

Coming Soon

Risk Assessment Framework

Framework for prioritizing and categorizing identified risks, with guidance on appropriate mitigation strategies.

Coming Soon

Research & Literature

Additional research and publications related to ML safety and FMEA methodologies.

Machine Learning Safety Survey

Faria, J. (2018)

Comprehensive survey of ML characteristics that safety engineers should understand, including potential failure modes in Markov Decision Processes and safe reinforcement learning.

View Paper

Taxonomy of Machine Learning Safety

Mohseni et al. (2022)

Links key safety principles to machine learning safety limitations and discusses three strategies: inherently safe design, enhancing performance & robustness, and runtime error detection.

View Paper

ML Safety in Automotive Software

Salay & Czarnecki (2018)

Assessment and adaptation of software process requirements in ISO 26262 for using machine learning safely in automotive applications.

View Paper

STPAI: STPA for AI Systems

Murphy (2024)

Application of Systems-Theoretic Process Analysis (STPA) to AI chatbot development, presented at the International Systems Safety Society 2024 conference.

View Paper

Community Resources

Connect with the ML FMEA community and access collaborative resources.

GitHub Discussions

Join discussions, ask questions, and share experiences with the ML FMEA community on GitHub.

Join Discussion

Community Forum

Dedicated forum for ML FMEA practitioners to share case studies, best practices, and implementation challenges.

Visit Forum

Events & Webinars

Stay updated on upcoming events, webinars, and workshops related to ML FMEA and AI safety.

View Events

Ready to Get Started?

Download the ML FMEA template and join our collaborative community to enhance AI safety.