Skip to content
mlmentorship

Book II · Chapter 1

Objectives and regularization

Start with training objectives, then control fit and generalization.

Priority
Core
Difficulty
Foundation
Useful for
All ML roles
Interview rounds
ML breadth, Training

Read first: Neural network foundations

Chapter contents

7 entries · read in order
  1. 01
    Cross-entropy and softmax
    Concept
  2. 02
    Regularization: L1, L2, dropout, early stopping, and the modern view
    Concept
  3. 03
    Dropout
    Concept
  4. 04
    Label smoothing
    Concept
  5. 05
    Weight decay vs. L2 regularization
    Concept
  6. 06
    Mixup and CutMix
    Concept
  7. 07
    Z-loss
    Concept