Familiarize yourself with relevant background

[maltemosbach]

First, familiarize yourself with the fundamentals that will be relevant to completing this project. The following resources should help you get started with reinforcement learning, deep learning, and state space models, but the usefulness will depend heavily on your prior knowledge. To better assess what will be beneficial, we can discuss your experience with the topics at our first meeting so that I can assess which resources and topics you should work through.

1. Reinforcement learning:

• Beginner's guide to RL
• SpinningUp in Deep RL ("Introduction to RL" section)
• UCL course on RL

2. Deep learning and PyTorch

• Beginner's guide to deep learning
• PyTorch basics tutorial

3. State space models, deep generative models, dynamics models for RL

• State space models
• Deep generative models ("Introduction" and "Variational Autoencoders" sections)
• PlaNet
• World models
• Auto-encoding variational bayes
• Dynamical variational autoencoders

[Resource]() = Further Reading. These resources are more detailed explanations that can be consulted for specific details or out of interest in a subject.

[karacolada]

Monday round-up:

• [x] completed PyTorch basics tutorial
• [x] skimmed Beginner's guide to Deep Learning (all known from lectures)
• [x] skimmed Beginner's guide to RL (Deep RL algos were unknown, rest known from lectures)
• [x] Deep RL parts 1+ 2

[karacolada]

Tuesday:

• [x] State space models
• [x] Deep generative models
• [x] World models
• [x] auto-encoding variational bayes