What is regularization in neural networks?
Updated May 17, 2026
Short answer
Regularization reduces overfitting by penalizing model complexity.
Deep explanation
Techniques include L1/L2 regularization, dropout, and early stopping. It constrains weight growth or model capacity.
Real-world example
Used in production ML models to improve generalization.
Common mistakes
- Over-regularizing leading to underfitting.
Follow-up questions
- What is L1 vs L2 regularization?
- What is early stopping?