The Neural Aesthetic @ ITP-NYU, Fall 2018
Lecture 2: Neural networks [9/11/2018]
[Slides]
- Admin (0:00)
- What are features (8:00)
- Neurons and linear functions (16:23)
- Non-linearities and hidden layers (20:23)
- Demo of a simple forward pass (24:49)
- ReLU activations (38:44)
- Neural net for classifying MNIST (40:19)
- Visualizing the weights (48:36)
- Visualizing weights of hidden layers (59:32)
- A more complicated dataset: CIFAR-10 (1:06:55)
- Precursors to convolutional networks (1:10:27)
- How convolutional layers work (1:22:55)
- Full demo of a convolutional network (1:32:37)
- Impact of convolutional networks (1:49:11)
- ml5.js image classifier (1:54:55)
- Transfer learning (2:03:26)
- ml5.js transfer learning demos (2:10:58)
- Transfer learning creative projects (2:20:34)