AI fundamentals · Question 29 of 60
How do generative adversarial networks (GAN) create realistic synthetic data?
1 answer is correct. Medium
Show answer
Answer
- C Two neural networks compete against each other
Explanation
A GAN consists of two neural networks that compete: one produces data, the other tries to expose it as fake. This results in very realistic synthetic data. Adding noise step by step and removing it again is the principle of diffusion models, and self-attention that of transformers. Many decision trees together form a random forest, which generates nothing.