Machine learning

A broad set of techniques and methods that allow AI systems to learn from large amounts of data, including supervised, unsupervised, and semi-supervised learning approaches.

Source: AIEOG AI Lexicon (Feb 2026), adapted from NIST AI 100-1

Machine learning (ML) is the subset of AI where systems learn from data rather than being explicitly programmed with rules. Instead of a developer writing code that says “if X, then Y,” machine learning algorithms discover patterns and relationships in data and use them to make predictions or decisions on new data.

Machine learning encompasses several learning approaches: supervised learning (learning from labeled examples), unsupervised learning (finding patterns in unlabeled data), semi-supervised learning (a mix of both), and reinforcement learning (learning through trial and error with feedback).

Machine learning is the dominant technology powering AI in financial services today. Credit scoring, fraud detection, transaction monitoring, customer segmentation, pricing optimization, and risk modeling all commonly use machine learning.

Machine learning is subject to model risk management requirements. SR 11-7, OCC guidance, and interagency statements all apply to ML models used for business decisions.

Key governance considerations include data dependency (ML models are only as good as their training data), continuous monitoring needs (ML models can drift as patterns change), validation complexity (ML models require specialized validation), and explainability challenges.

1. Apply model risk management. All ML models used for business decisions should be subject to your model risk management framework.
2. Validate rigorously. ML models require validation that assesses performance, bias, stability, and sensitivity to data changes.
3. Monitor continuously. Implement ongoing performance monitoring with defined thresholds and escalation procedures.
4. Ensure data quality. Assess and document the quality of training data and production data feeding ML models.
5. Document model design. Maintain documentation of the ML technique chosen, alternatives considered, feature selection, and training approach.
6. Plan for retraining. Establish defined procedures for when and how ML models will be retrained as data patterns evolve.

Deep learning, AI model, Supervised learning, Unsupervised learning, Reinforcement learning, Training data, Algorithm, AI lifecycle