Virtual Labs

Generative Adversarial Networks (GANs)

1. What is the primary application of Deep Convolutional GAN (DCGAN)?

a: Image-to-image translation Explanation

Explanation

b: Style transfer Explanation

Explanation

c: High-resolution image generation Explanation

Explanation

d: Anomaly detection Explanation

Explanation

2. Why is adversarial training used in GANs?

a: To train both networks independently Explanation

Explanation

b: To create competition between Generator and Discriminator Explanation

Explanation

c: To speed up the training process Explanation

Explanation

d: To eliminate the need for labelled data Explanation

Explanation

3. What is the purpose of noise input in the generator component of a GAN?

a: To randomize the generation process and introduce variations Explanation

Explanation

b: To control the learning rate during training Explanation

Explanation

c: To adjust the weights and biases of the generator Explanation

Explanation

d: None of the above Explanation

Explanation

4. What does the discriminator output contribute to during GAN training?

a: Provide feedback to the generator and help it improve Explanation

Explanation

b: Generate synthetic samples Explanation

Explanation

c: Adjust the learning rate during training Explanation

Explanation

d: None of the above Explanation

Explanation

5. What does the term "adversarial" in GANs refer to?

a: The competitive process between two sub-models Explanation

Explanation

b: The generation of new data Explanation

Explanation

c: The architecture of the generator Explanation

Explanation

d: The classification of real data Explanation

Explanation

6. What type of neural network layers are commonly used in DCGAN?

a: Recurrent layers Explanation

Explanation

b: Fully connected layers Explanation

Explanation

c: Convolutional layers Explanation

Explanation

d: Max-pooling layers Explanation

Explanation

7. Which optimizer was used in the original DCGAN paper?

a: Gradient Descent Explanation

Explanation

b: Adam Explanation

Explanation

c: RMSprop Explanation

Explanation

d: AdaGrad Explanation

Explanation

8. Why are convolutional layers used in DCGAN instead of fully connected layers?

a: They require fewer parameters and capture spatial features Explanation

Explanation

b: They guarantee perfect image generation Explanation

Explanation

c: They eliminate the need for a discriminator Explanation

Explanation

d: They work only for grayscale images Explanation

Explanation

9. What does the "Real vs Generated" panel help in understanding?

a: Numerical accuracy of the model Explanation

Explanation

b: Training speed of the optimizer Explanation

Explanation

c: Visual quality and realism of generated images Explanation

Explanation

d: Memory usage of the GPU Explanation

Explanation

10. What is the format and nature of images in the MNIST dataset used for GAN training?

a: RGB images of size 224×224 Explanation

Explanation

b: Grayscale images of handwritten digits Explanation

Explanation

c: Colour images with multiple objects Explanation

Explanation

d: Text-based image representations Explanation

Explanation