1. Question 1 — What is convergence?

• ❌ A programming API for AI

• ❌ A dramatic increase in loss

• ❌ Overfitting definition

• ✅ The process of getting very close to the correct answer

Explanation:
Convergence means the model is approaching the minimum of the loss function.

---

2. Question 2 — Difference between traditional programming & ML

• ✅ In traditional programming, rules are manually coded; in ML, the algorithm learns rules from data.

• ❌ Activity recognition difference

---

3. Question 3 — What does model.fit() do?

• ✅ It trains the neural network to fit inputs (X) to expected outputs (Y)

• ❌ Fits available memory

• ❌ Just optimizes

• ❌ Determines if activity is good

---

4. Question 4 — What do we call the process of telling the computer what the data represents?

• ❌ Programming the Data

• ❌ Learning the Data

• ❌ Categorizing the Data

• ✅ Labeling the Data

---

5. Question 5 — What does the optimizer do?

• ❌ Stops training

• ✅ Updates weights to reduce loss

• ❌ Compiles code

• ❌ Measures guess quality (that’s the loss function)

---

6. Question 6 — What is a Dense layer?

• ❌ Definition of density

• ❌ Disconnected neurons

• ❌ Single neuron

• ✅ Layer of neurons fully connected to adjacent layers

---

7. Question 7 — How do we measure how good the current guess is?

• ❌ Win/lose

• ✅ Using the loss function

• ❌ Training the NN

---

8. Question 8 — How to define input shape?

• ❌ No need

• ✅ Using a tf.keras.Input with shape argument

• ❌ InputLayer

• ❌ input_shape in Dense (works, but not best practice)

---

🧾 Summary Table

---