Generative AI

Question 1

Diffusion Models

Answer 1

Markov Chain: Models sequence of noise distributions
Generates by de-noising
Forward diffusion: Introduce noise (gaussian) into the input until it becomes completely noisy. Markov Chain, doesn’t require training
Backward diffusion: Denoising process learned with NN

Question 2

GANs

Question 3

VAEs

Question 4

What is diffusion in physics?

Answer 4

Movement of particles from high concentration to low concentration
Increases entropy

Question 5

Markov Chain

Answer 5

A stochastic model describing a sequence of events where the probability of each event depends only onto the previous state

Question 6

Monte Carlo Simulations

Answer 6

Involve repeated random samplings

Question 7

Markov Chain Monte Carlo Sampling