Virtual Labs

Perceptron

1. When does a perceptron update its weights during training?

a: After every epoch, regardless of accuracy Explanation

Explanation

b: Only when it makes an incorrect prediction Explanation

Explanation

c: Before making any predictions Explanation

Explanation

d: Only in the first epoch Explanation

Explanation

2. What is the first step a perceptron takes when making a prediction?

a: Calculate the weighted sum of inputs plus bias Explanation

Explanation

b: Apply the activation function Explanation

Explanation

c: Update the weights Explanation

Explanation

d: Generate random outputs Explanation

Explanation

3. What output does the perceptron's step function produce when z = 0?

a: 0.5 Explanation

Explanation

b: -1 Explanation

Explanation

c: Undefined Explanation

Explanation

d: 1 Explanation

Explanation

4. If you observe the decision boundary after each epoch, what would you notice?

a: It rotates and shifts position trying to separate the classes Explanation

Explanation

b: It stays in the same position throughout training Explanation

Explanation

c: It splits into multiple lines Explanation

Explanation

d: It becomes a curve Explanation

Explanation

5. Which pairs of XOR input points belong to the same class label?

a: (0,0) and (0,1) Explanation

Explanation

b: (1,0) and (1,1) Explanation

Explanation

c: (0,0) and (1,1), or (0,1) and (1,0) Explanation

Explanation

d: All four points share the same label Explanation

Explanation

6. What does the error value represent in the perceptron learning rule?

a: The difference between actual and predicted labels Explanation

Explanation

b: The learning rate Explanation

Explanation

c: The total number of training epochs Explanation

Explanation

d: The weight value Explanation

Explanation

7. When does the perceptron learning algorithm update the bias term?

a: Only at the end of each epoch Explanation

Explanation

b: The bias never changes during training Explanation

Explanation

c: Only when accuracy drops below 50% Explanation

Explanation

d: When there is a prediction error, using: b = b + learning_rate × error Explanation

Explanation

8. What is the fundamental lesson learned from the XOR experiment?

a: Perceptrons are useless for all problems Explanation

Explanation

b: More training epochs will eventually solve any problem Explanation

Explanation

c: Single layer perceptrons have a fundamental limitation in solving non-linearly separable problems Explanation

Explanation

d: Learning rate is the most important hyperparameter Explanation

Explanation

9. Which components of a perceptron are adjusted during the training process?

a: Both weights and bias Explanation

Explanation

b: Only the weights Explanation

Explanation

c: Only the input values Explanation

Explanation

d: Only the activation function Explanation

Explanation

10. What is an "epoch" in the context of perceptron training?

a: A single prediction made by the perceptron Explanation

Explanation

b: One complete pass through the entire training dataset Explanation

Explanation

c: The final accuracy of the model Explanation

Explanation

d: The time taken to train the model Explanation

Explanation