🧪 05 — Exercises: Deep Learning

Tasks

• explain neuron, layer, activation, loss, optimizer
• build a small Keras model
• train for 20 epochs
• plot training vs validation loss
• add early stopping
• compare to Logistic Regression or Random Forest

---

Interview Questions

Q1: What is backpropagation?

The algorithm that computes gradients so weights can be updated to reduce loss.

Q2: What is overfitting?

When a model performs well on training data but poorly on unseen data.

---

Next

➡️ NLP Basics