感觉是理论和应用都很好地结合的一门课，理论方面扎实而不死板，需要关于概率和信息论的基础，project和课上用jupyter notebook举的例子都很有利于之后的实践。
关于课程内容, 分享一下最后一节课ppt的内容吧
"""You’ve learned a lot!
Bayesian linear regression, Gaussian processes, Bayesian deep learning, variational inference, MCMC, SGLD, Gibbs sampling, bandits, Bayesian optimization, Markov Decision processes, value iteration, policy iteration, POMDPs, TD-learning, Qlearning, DQN, actor-critic methods, model-based deep reinforcement learning, Bayesian RL, PETS, H-UCRL, …
"""
"""Key concepts & notions
- Bayesian learning
- Learning as inference
- Epistemic vs aleatoric uncertainty
- Score- and reparametrization gradient estimators
- POMDPs as belief-state MDPs
- Optimism in the face of uncertainty
"""