1️⃣ Binary logistic regression §

= 🚶 Logistic Regression for binary classification

Phases §

Training §

1. Calculate the difference between the classifier output and the true value
   • → Loss function: Cross-entropy loss
2. Optimize the weights to minimize the loss
   • → Optimization algorithm: Stochastic gradient descent

How much training? §

• Avoid overfitting: → Regularization by adding $R(\theta )$ to the loss function, which penalizes large weights
• Model should be able to generalize

Test §

= testing whether a document belongs to a class

1. Calculate the Z-Score of the feature vector
2. Use the Logistic Sigmoid Function to turn it into a 🎲 Probability