Overfitting is a phenomenon in which a model learns the training data so literally—including its noise and random outliers—that it loses the ability to generalize correctly to new, previously unseen data. It manifests as excellent results on the training set but significantly poorer performance in real-world applications, and is one of the main risks in the development of any AI model.
Glossary
Overfitting
Overfitting
Similar articles
View all articles
Academy Article
13. 9. 2026
2026 Retraining Grant: How to Get a Free Course Through the Employment Office (Complete Guide)
14 min read
Academy Article
13. 9. 2026
Where Companies Will Save the Most with AI Automation in 2026 (and Where They’re Wasting Money)
8 min read
Academy Article
13. 9. 2026