6.2 - Machine Learning - Key Concepts & Supervised ML : Linear Regression

Week 6

Objectives

• ML Terminology:
  • Labels, Features, Examples, Models
  • Training, Inference/Testing,
• Linear Regression:
  • Equation, Loss, Update weights based on Loss
  • LASSO, RIDGE
• Gradient Descent
  • Stochastic, Batch, Mini-batch stochastic
• Loss Functions :
  • Esp MAE, MSE
• Learning rate
• Overfitting v/s Underfitting
• Data -
  • Split as (Train, Val, test)
  • K Cross validation
• Regularisation :
  • Reducing Model Complexity:
    • L1/L2 Regularisation
    • Dropout
    • Early stopping
  • Data Augmentation

Materials

• Slides
• Board I used

Lecture

Next Class

6.3 Supervised Learning - Logistic Regression & Getting to code