Deep Learning Generative Models 2 — Questions and Answers
Question 1: What is an autoencoder and what are its two main components?
- A generative model; generator and discriminator
- A neural network that compresses input to a low-dimensional latent code and reconstructs it; encoder and decoder (Correct answer)
- A sequence model; embedding layer and recurrent layer
- A supervised classifier; feature extractor and classification head
Correct answer: A neural network that compresses input to a low-dimensional latent code and reconstructs it; encoder and decoder
An autoencoder maps input to a compressed latent representation via the encoder and reconstructs the input from that representation via the decoder, learning compact data representations.
Question 2: What is the Wasserstein GAN (WGAN) designed to address?
- Mode collapse in image synthesis
- Training instability and mode collapse by using the Wasserstein distance as a more stable loss (Correct answer)
- Slow convergence in the discriminator
- Class-conditional generation without labels
Correct answer: Training instability and mode collapse by using the Wasserstein distance as a more stable loss
WGAN replaces the standard GAN loss with the Wasserstein distance (Earth Mover's distance), providing more meaningful gradients even when generator and real distributions don't overlap.
Question 3: What is conditional generation in generative models?
- Generating samples only when the model has high confidence
- Conditioning the generative process on a label or other input to control the attributes of generated samples (Correct answer)
- Training the model only on samples satisfying a quality threshold
- Applying post-processing conditions to raw generated outputs
Correct answer: Conditioning the generative process on a label or other input to control the attributes of generated samples
Conditional generation feeds class labels, text prompts, or other conditioning signals to the generative model, enabling control over the category or attributes of generated outputs.
Question 4: What is latent space interpolation in generative models?
- Mixing two training images at the pixel level
- Smoothly traversing the latent space between two encoded points to generate intermediate samples (Correct answer)
- Randomly sampling the latent space to generate diverse outputs
- Extrapolating beyond the training distribution to novel examples
Correct answer: Smoothly traversing the latent space between two encoded points to generate intermediate samples
Latent space interpolation linearly or spherically blends two latent vectors and decodes the intermediate points, producing samples that smoothly transition between the two originals.
Question 5: What is the key property of flow-based generative models such as RealNVP?
- They use adversarial training between two networks
- They learn invertible transformations allowing exact likelihood computation and efficient sampling in both directions (Correct answer)
- They encode inputs to discrete latent codes
- They model the data distribution using a recurrent prior
Correct answer: They learn invertible transformations allowing exact likelihood computation and efficient sampling in both directions
Flow-based models use a series of invertible (bijective) transformations with tractable Jacobians, enabling exact log-likelihood evaluation and straightforward sampling by inverting the flow.
Question 6: What is text-to-image generation and which model type is currently dominant for this task?
- Converting image captions to text descriptions; RNNs
- Generating photorealistic images from natural language descriptions; diffusion models conditioned on text embeddings (Correct answer)
- Transcribing handwritten text in images; CNNs
- Retrieving images from a database matching a text query; retrieval models
Correct answer: Generating photorealistic images from natural language descriptions; diffusion models conditioned on text embeddings
Text-to-image generation produces images matching a text prompt; diffusion models conditioned on CLIP or text encoder embeddings are the current state-of-the-art approach.
Deep Learning Generative Models 2
What is an autoencoder and what are its two main components?