Virtual Labs

Recurrent Neural Networks (RNN)

1. Which one of the following tasks is most suitable for RNNs?

a: Image classification Explanation

Explanation

b: Text generation Explanation

Explanation

c: Reinforcement learning Explanation

Explanation

d: Object detection Explanation

Explanation

2. What is the main characteristic of a Recurrent Neural Network (RNN)?

a: It processes data sequentially and has loops in its architecture Explanation

Explanation

b: It processes data in parallel and has no loops Explanation

Explanation

c: It uses a single-layer architecture Explanation

Explanation

d: It is designed for image classification Explanation

Explanation

3. Which layer type is typically used to capture sequential dependencies in an RNN?

a: Input layer Explanation

Explanation

b: Hidden layer Explanation

Explanation

c: Output layer Explanation

Explanation

d: Activation layer Explanation

Explanation

4. What is the advantage of using recurrent layers in an RNN?

a: They can capture temporal dependencies in the input data Explanation

Explanation

b: They process all inputs simultaneously in parallel Explanation

Explanation

c: They can generate synthetic data Explanation

Explanation

d: They can handle non-linear transformations Explanation

Explanation

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

a: To store the information from the previous time step Explanation

Explanation

b: To adjust the learning rate during training Explanation

Explanation

c: To compute the gradients for backpropagation Explanation

Explanation

d: None of the above Explanation

Explanation

6. Why do Recurrent Neural Networks (RNNs) share parameters across time steps?

a: To allow the model to learn different features at each step Explanation

Explanation

b: To increase the total number of learnable parameters Explanation

Explanation

c: To enable the network to generalize across sequences of varying lengths Explanation

Explanation

d: To eliminate the need for an activation function Explanation

Explanation

7. What does the number of time steps in an unrolled RNN correspond to?

a: The number of hidden layers in the network Explanation

Explanation

b: The total number of parameters in the network Explanation

Explanation

c: The length of the input sequence being processed Explanation

Explanation

d: The number of output classes Explanation

Explanation

8. What is a common challenge when training standard Recurrent Neural Networks on very long sequences?

a: The network learns to process data too quickly Explanation

Explanation

b: The vanishing or exploding gradient problem Explanation

Explanation

c: The inability to use any activation functions Explanation

Explanation

d: The requirement to have a fixed input size Explanation

Explanation

9. What is the main difference between an RNN and a feedforward neural network?

a: RNN uses convolution layers Explanation

Explanation

b: RNN has feedback connections Explanation

Explanation

c: RNN has no hidden layers Explanation

Explanation

d: RNN works only on images Explanation

Explanation

10. What dataset is commonly used for character-level language modeling demonstrations?

a: MNIST Explanation

Explanation

b: CIFAR-10 Explanation

Explanation

c: ImageNet Explanation

Explanation

d: Tiny Shakespeare Explanation

Explanation