Generative Adversarial Networks(Generative AI) QUIZ (MCQ QUESTIONS AND ANSWERS)

The competition between multiple discriminators

The collaboration between generator and discriminator

The attempt to minimize the generator's loss function

The adversarial relationship between the generator and discriminator

Discriminator

Encoder

Generator

Classifier

Dropout

L1 Regularization

L2 Regularization

Gradient Penalty

Learning rate decay

Learning rate annealing

Learning rate scheduling

Learning rate adjustment

Limited scalability to large datasets

High computational cost

Difficulty in training stable models

Inability to generate high-resolution images

Early stage

Middle stage

Late stage

Throughout the training process

Feature matching

Minibatch discrimination

Gradient penalty

Spectral normalization

Limited training stability

Longer training time

Higher computational cost

Difficulty in controlling mode collapse

Conditional GAN (cGAN)

Deep Convolutional GAN (DCGAN)

Wasserstein GAN (WGAN)

InfoGAN (Information Maximizing GAN)

By generating synthetic anomalies for training

By classifying anomalies in existing data

By compressing anomaly data

By ignoring anomaly detection tasks

Limited scalability to large datasets

High computational cost

Difficulty in training stable models

Inability to generate high-resolution images

Diversity regularization

Data augmentation

Latent space interpolation

Minibatch discrimination

Gradient Descent

Adam

RMSprop

AdaGrad

Learning rate decay

Learning rate annealing

Learning rate scheduling

Learning rate adjustment

Skip connections

Stochastic gradient descent

Batch Normalization

Feature matching

Image classification

Image generation

Image compression

Image segmentation

Early stage

Middle stage

Late stage

Throughout the training process

Mode collapse

Vanishing gradients

Exploding gradients

Training instability

Improved training stability

Faster convergence

Lower computational cost

Better scalability

DCGAN (Deep Convolutional GAN)

WGAN (Wasserstein GAN)

InfoGAN (Information Maximizing GAN)

CycleGAN

Image classification

Image generation

Speech recognition

Natural language processing

Unsupervised learning

Supervised learning

Semi-supervised learning

Reinforcement learning

To minimize the loss function of the Discriminator

To maximize the loss function of the Discriminator

To minimize the loss function of the Generator

To maximize the loss function of the Generator

To minimize the loss function of the Discriminator

To maximize the loss function of the Discriminator

To minimize the loss function of the Generator

To maximize the loss function of the Generator

Cross-Entropy Loss

Mean Squared Error (MSE) Loss

Binary Cross-Entropy Loss

Kullback-Leibler (KL) Divergence

Cross-Entropy Loss

Mean Squared Error (MSE) Loss

Binary Cross-Entropy Loss

Kullback-Leibler (KL) Divergence

To generate fake data samples

To evaluate the authenticity of data samples

To classify data samples

To preprocess input data

To generate fake data samples

To evaluate the authenticity of data samples

To classify data samples

To preprocess input data

Encoder and Decoder

Generator and Discriminator

Classifier and Regressor

Reinforcement Learning Agent and Environment

Andrew Ng

Yann LeCun

Ian Goodfellow

Geoffrey Hinton