Which of the following is a primary component of a GAN architecture?

• Encoder
  

•  Decoder
  

• Generator
  

• Feature Mapper
  

What is the main objective of the Discriminator in a GAN?

• Generate realistic data

• Distinguish between real and generated data
  

• Train the Generator using labels
  

• Reduce model size
  

Which ensemble method builds multiple models in parallel to reduce variance in predictions?

•  Stacking

• Boosting

• Bagging
  

• Blending

The latent representation in a Variational Autoencoder is used for:

• Calculating classification accuracy
  

• Encoding compressed data representations
  

• Saving training checkpoints
  

• Preprocessing input images
  

CycleGANs are best used when:

• The dataset is labeled with pairs of corresponding images
  

• There are unpaired datasets for image translation
  

•  Only textual data is present
  

•  Training is supervised with explicit targets
  

In Gaussian Mixture Models (GMMs), which parameter determines the spread of each Gaussian component?

• Mean vector

• Mixing coefficient

• Weight vector

• Covariance matrix

Which type of autoencoder is specifically designed to handle noisy input data by learning robust representations?

• Sparse autoencoder

• Undercomplete autoencoder

• Denoising autoencoder

• Contractive autoencoder

What is the primary advantage of using XGBoost over traditional gradient boosting methods?

• Parallel processing capability

• Smaller model size

• Fewer hyperparameters

• Linear time complexity

In transfer learning, what is feature extraction primarily used for?

• Creating new features from scratch

• Using pre-trained model layers without modification

• Removing unnecessary features

• Transforming input data format

Which component of the Metaverse ecosystem is most crucial for creating immersive AI experiences?

• Blockchain integration

• Social networking features

• Virtual/Augmented reality interfaces

• Digital currency support