A student-run course I taught at UVA for upper class computer science students. Student-run courses are only allowed to be one-credit, which means that there was little opportunity for discussing code. Instead, the focus was on the mathematical foundation of neural networks and how the model can be extended to achieve desired results.
The course was split up into three sections:
- An introduction to machine learning and neural networks for four lectures.
- After this, specific types of neural networks were discussed.
- These are auto-encoders, Convolutional networks and Recurrent Networks/LSTMs.
- Then, we read six papers that are important in the field or just really cool. The papers in order were:
- AlexNet and ZFNet (both papers discussed in one lecture)
- Dynamic Pooling and Unfolding Recursive Autoencoders for Paraphrase Detection
- Pixel Recursive Super Resolution
- Mastering the Game of Go with Deep Neural Networks and Tree Search
- GANs