Gan Quiz

What does GAN stand for?

• Generative Adversarial Network
• Graph Attention Network
• Gaussian Adaptive Network
• General Adaptive Node

Answer explanation

GAN stands for Generative Adversarial Network, a class of machine learning models used for generative tasks.

GANs are primarily used for which of the following tasks?

• Regression analysis
• Time series prediction
• Generative tasks such as image synthesis and style transfer
• Classification only

Answer explanation

GANs are widely used for generative tasks including image synthesis, style transfer, and data augmentation.

What is the main function of the Generator in a GAN?

• To evaluate data authenticity
• To compress data
• To take random noise and generate realistic-looking samples
• To classify images into categories

Answer explanation

The generator takes random noise as input and generates samples that ideally resemble the training data.

What is the primary role of the Discriminator in a GAN?

• To perform feature extraction
• To evaluate whether input is real or generated
• To reduce model complexity
• To generate synthetic data

Answer explanation

The discriminator is a binary classifier that evaluates whether a given input is real from the training set or fake from the generator.

How are the Generator and Discriminator trained in a GAN?

• Only the discriminator is trained
• Independently without interaction
• Only the generator is trained
• Simultaneously in an adversarial manner

Answer explanation

The generator and discriminator are trained simultaneously in a competitive manner where each aims to outsmart the other.

What type of input does the Generator typically receive?

• Class labels only
• Raw pixel values
• Random noise vector
• Labeled training images

Answer explanation

The generator takes random noise as input, often sampled from a Gaussian or uniform distribution.

Which of the following is a known challenge in training GANs?

• Lack of gradient flow
• Too many training samples
• Mode collapse and training instability
• Convergence is too fast

Answer explanation

Training GANs can be challenging and requires careful tuning of hyperparameters. Mode collapse is a common issue where the generator produces limited variety.

What is one practical application of GANs mentioned in the presentation?

• Spam detection
• Password encryption
• Image synthesis and deepfake video production
• Email filtering

Answer explanation

GANs have been successful in generating high-quality images, creating art, and producing deepfake videos.

What does the Discriminator output in a GAN?

• A reconstructed image
• A feature vector
• A probability score indicating if input is real or fake
• A class label

Answer explanation

The discriminator acts as a binary classifier, outputting a probability that indicates whether the input is real or generated.

Besides GANs, which other generative technique is mentioned as a key feature in Generative AI?

• Random Forests
• Support Vector Machines
• Variational Autoencoders (VAE)
• Decision Trees

Answer explanation

The presentation mentions both GANs and Variational Autoencoders (VAE) as key features in Generative AI.