Overfitting
A modelling problem where a machine learning model learns the specific patterns and noise in training data too closely, resulting in poor generalisation to new, unseen data.
In plain language
When an AI memorises the training data rather than learning general patterns that work on new data. Like a student memorising test answers instead of learning concepts; they ace the test but fail on a new exam.
Why this matters
Overfitting is a critical risk because models may perform excellently during testing but fail in production on real data. Your AI governance framework must require rigorous validation on independent test sets and ongoing monitoring in production to detect overfitting before it causes business or safety harm.
Relevance
ImplementationRequires robust validation practices and ongoing monitoring to detect performance degradation in production, which is essential for maintaining model reliability.
Related terms
Putting overfitting into practice in your organisation?
Ready to transform your AI strategy?
Partner with Australia's AI strategy and governance specialists. From adoption roadmaps to ISO 42001 audit readiness.