In [1]:
from IPython.display import display, Markdown

display(Markdown(content))


# Bayes_By_Example¶

What is this course about? I recently started a reinforcement learning project where the reward function of the RL learner was a bayesian model. Of course...I knew almost nothing about bayesian statistics before building out this RL learner! So, I spent a lot of time reading and trying to implement various online tutorials and research papers. This course is a collection of (mostly) other people's tutorials, but refactored to work together and provide a more holistic view of bayesian statistics than any one of these myriad of tutorials did originally. I've rewritten extensive portions of many of these tutorials and I apologize for any errors I may have accidentally introduced. I try to provide links or references to any and all resources I borrowed from to create this course in the sections where I reference said material.

All of the content is stored in Jupyter Notebooks. I recommend downloading the content locally. This will let you play with the code and actually run the examples for yourself. For best practices, I also recommend using a conda virtual environment to manage dependancies (take a look at the setup.sh file if you need help setting up your env and/or check out the documentation).

Part One:

• Who is this lecture for?
• What is statistics?
• The Frequentist Philosophy
• The Bayesian Philosophy
• Author's Note: Taking Sides

Part Two:

• Bayes: Conjugacy
• Bayes: Metropolis Hastings Sampling
• Bayes: My Friends Edward, Stan, and Pymc3

Part Three:

• Bayes: Hyper Priors
• Bayesians vs Frequentists (Student's T Test)
• Bayesians vs Frequentists (A/B Testing)
• Bayes: Gibbs Sampling

Part Four:

• Bayes: Linear Regression Part One (simple)
• Bayes: Linear Regression Part Two (regularization)
• Bayes: Linear Regression Part Three (robust)

Part Five:

• Bayes: Linear Regression Four (multi-level)
• Bayes: Linear Regression Four and Three Quarters (escaping the funnel)
• Samplers: Beyond Metropolis Hastings

Part Six:

• Bayes: Linear Regression with a Time Component (includes neural net with time component)

Part Seven:

• Bayes: Variational Inference

Part Eight:

• Bayes: Training Neural Nets Part One
• Bayes: Training Neural Nets Part Two
• Bayes: Training Neural Nets Part Three
• Bayes: Reguarization
• Bayes: Time Series Neural Nets

Part Nine:

• Mixture Models Part One
• Mixture Models Part Two
• Mixture Models Part Three
• Mixture Models Part Four

Part Ten:

• Gaussian Processes
• Parting Words

Part Eleven:

• Naive Bayes
In [ ]: