The Neural Aesthetic @ ITP-NYU, Fall 2018
Lecture 6: Generative models [10/23/2018]
[Slides]
- Why study generative models? (3:29)
- What are generative models? (8:26)
- Dimensionality reduction and PCA demo (12:15)
- Pixel-space and the curse of dimensionality (21:57)
- Eigenfaces (29:46)
- Linear PCA vs Non-linear methods (45:25)
- Neural net & embeddings review (51:37)
- Autoencoders (56:26)
- Generative adversarial networks (1:10:04)
- DCGANs and feature arithmetic (1:15:30)
- DCGAN examples projects (1:19:35)
- Deep generator networks (1:31:19)
- High-resolution and progressively-grown GANs (1:34:37)
- GLOW and reversibility, fMRI-conditioned GANs (1:40:53)
- Generative models in text and audio domain (1:45:47)
- Practical resources and tutorials (1:53:39)
- Scraping Instagram, Google, and Bing (1:55:00)
- Finding publicly available datasets (2:03:42)
- Dataset utils for pre-processing datasets (2:07:35)
- Setting up Paperspace job-runner for DCGAN (2:18:09)
- Training DCGAN-tensorflow (2:27:07)
- Other GAN-training resources (2:38:46)