1. GAN
GAN models are known for potentially unstable training and less diversity in generation due to their adversarial training nature.
2. VAE
VAE relies on a surrogate loss.
- 如何简单易懂地理解变分推断(variational inference)？
- Inference Suboptimality in Variational Autoencoders
- The Reparameterization Trick
- Paper List
3. Flow-based Model
Flow models have to use specialized architectures to construct reversible transform.
4. Diffusion Model
