DecodeAI
Machine Learning Foundations
Pro topic

Regularization & Generalization

Bias-variance, early stopping, and capacity control.

Regularization & Generalization

Overfitting means low training error but poor unseen performance.

Levers

  • More / cleaner data
  • Simpler hypothesis class
  • L2 / weight decay, dropout, early stopping
  • Data augmentation (when valid)

Bias–variance intuition

High bias: model too simple. High variance: model too sensitive to sample noise. Regularization trades a bit of bias for lower variance.

Continue this theory with Pro

You've reached the free preview. Pro unlocks the full write-up, videos, quiz scoring, and cheatsheet.