Machine Learning Assurance (MLA) is a controls-based process for ML systems that establishes confidence and verifiability through software and human oversight.
The objective of MLA is to assure interested stakeholders that an ML system is functioning as expected, but in particular, to assure an ML system's transparency, compliance, fairness, safety, and optimal operation.
ML systems present a new paradigm of technology and business decisioning.
Individuals and organizations responsible for managing risk and compliance – including regulators – have not previously contended with such dynamic and variable models and systems.
Huge data collections
Rapid, opaque decisions
Evolving, rapidly scaling models
Specialized, complex technology
Because machine learning is making key decisions that affect people's lives, livelihoods, and opportunities, trust and confidence in ML systems is paramount. There should be a reasonable ability to evaluate and verify their safety, fairness, and compliance.
Companies, regulators, and consumers all benefit from an objective method of assuring ML systems.
ML models can behave unexpectedly, creating enormous liabilities and risk for unforeseen damages and reputational harm.
ML systems are opaque to non-technical professionals, requiring more ongoing attention from objective parties to ensure safety and compliance.
Unassured ML applications can lead to broad misconceptions and bias, undermining the long-term promise of the technology.
Three core principles empower an effective MLA function and responsible use of ML.
MLA requires clear understanding and documentation of the considerations, goals, and risks evaluated during the lifecycle of an ML application.
MLA requires each business and technical decision and step have the ability to be verified and interrogated.
MLA requires any ML application can be reasonably evaluated and understood by an objective individual or party not involved in the model development.
Evaluate key business drivers and questions at project inception and revisit at every following step.
Understand data lineage, quality, and usage rights.
Control: Data governance
Pre-process standardized data sets for training, test, and production.
Controls: Data Preparation, Data Segmentation
Create the simplest, best fit, and performant models.
Controls: Algorithm Selection, Cross-Functional Review, Metric Selection
Determine accuracy and precision of models before launch and continuously in production.
Control: Model Validation
Deliver business sign-off and develop capacity for continuous inspection and review.
Controls: Executive Accountability, Monitoring Process, Model Logging & Interpretability
Below is a list of organizations and websites helping to drive the conversation about Machine Learning Assurance.
ISACA: The Machine Learning Audit – CRISP-DM Framework
ICO: AI Auditing Framework
ICO: Guidance on the AI auditing framework
From our founding, we've focused entirely on Machine Learning Assurance. We actively collaborate with commercial, regulatory, and standards organizations to support education and adoption of MLA. Join us in our mission to create trust and confidence in machine learning.