Virtual Labs

1. What is the purpose of the recurrent connection in an RNN?

a: To propagate the hidden state across different time steps Explanation

Explanation

b: To adjust the weights and biases Explanation

Explanation

c: To reduce the dimensionality of the input Explanation

Explanation

d: None of the above Explanation

Explanation

2. What is the purpose of the initial hidden state in an RNN?

a: To provide the starting point for the recurrent computation Explanation

Explanation

b: To adjust the learning rate Explanation

Explanation

c: To compute gradients Explanation

Explanation

d: None of the above Explanation

Explanation

3. How does the RNN handle sequential data?

a: Using a sliding window Explanation

Explanation

b: Processing data point by point with memory of past inputs Explanation

Explanation

c: Averaging inputs Explanation

Explanation

d: Applying convolution Explanation

Explanation

4. In the context of RNNs, what does "unrolling" refer to?

a: Training with large batch size Explanation

Explanation

b: Unfolding the RNN across time steps Explanation

Explanation

c: Reducing number of layers Explanation

Explanation

d: Implementing dropout Explanation

Explanation

5. What is the primary reason for visualizing sequence unrolling in RNNs?

a: To reduce model complexity Explanation

Explanation

b: To understand how parameters change Explanation

Explanation

c: To visualize how information flows over time Explanation

Explanation

d: To improve accuracy directly Explanation

Explanation

6. What happens if the initial hidden state of an RNN is set to zero?

a: The model cannot learn Explanation

Explanation

b: The network starts without prior memory Explanation

Explanation

c: The first output is ignored Explanation

Explanation

d: The gradients become zero permanently Explanation

Explanation

7. Which visualization helps understand memory behaviour in RNNs?

a: Accuracy curve Explanation

Explanation

b: Confusion matrix Explanation

Explanation

c: ROC curve Explanation

Explanation

d: Hidden-state evolution Explanation

Explanation

8. During text generation, what is used as input after the first character?

a: Random noise Explanation

Explanation

b: Original training data Explanation

Explanation

c: Previously generated character Explanation

Explanation

d: Entire sentence Explanation

Explanation

9. What technique is used to train RNNs over multiple time steps?

a: Forward propagation Explanation

Explanation

b: Gradient descent Explanation

Explanation

c: Dropout Explanation

Explanation

d: Backpropagation Through Time (BPTT) Explanation

Explanation

10. What problem commonly occurs in vanilla RNNs when handling long sequences?

a: Overfitting Explanation

Explanation

b: Exploding memory Explanation

Explanation

c: Vanishing gradients Explanation

Explanation

d: Data leakage Explanation

Explanation

Recurrent Neural Networks (RNN)