GAN Variants MCQ on 18.2.2025

What is the main purpose of Conditional GANs (CGANs)?

• The main purpose of Conditional GANs (CGANs) is to generate data conditioned on specific input information.

• To enhance the quality of existing data.

• To reduce the dimensionality of data.

• To classify data into predefined categories.

Describe the key components of the DCGAN architecture.

• Fully connected layers and dropout layers

• Key components of the DCGAN architecture are the generator, discriminator, transposed convolutional layers, convolutional layers, batch normalization, and Leaky ReLU activation functions.

• Pooling layers and sigmoid activation functions

• Recurrent layers and attention mechanisms

What is the primary advantage of using the WGAN loss function?

• Stability in training and improved gradient flow.

• Faster convergence and reduced training time.

• Increased likelihood of mode collapse.

• Higher model complexity and capacity.

List two common challenges faced during GAN training.

• Overfitting and underfitting

• High computational cost and slow convergence

• Lack of diversity in training data

• Mode collapse and instability in training.

How do Conditional GANs differ from standard GANs?

• Conditional GANs differ from standard GANs by conditioning the generation process on additional information, allowing for more controlled output.

• Conditional GANs generate outputs without any constraints.

• Conditional GANs do not require any additional input data.

• Conditional GANs use a different loss function than standard GANs.