Start Here: How to Read This Book

Welcome! This book teaches probability, probabilistic machine learning, and probabilistic computing through one continuous story: a school year in the life of Chibany, the Chiba Tech mascot, who receives two bentos a day from students and refuses to stop asking questions about them.

It was created with designers and social scientists in mind โ€” no prior math background is required. Every idea arrives in three steps: a concrete scene first, the math second, runnable GenJAX code third.

Choose your track

There are two guided paths through the book. Pick one with the ๐Ÿงญ Reading track switch in the sidebar (your choice is remembered); off-track chapters stay visible but dimmed, and each chapter’s footer shows the next stop on your path.

๐Ÿค– Beginner

You want to learn probability theory, probabilistic machine learning, and probabilistic computing from a light background โ€” through to neural networks, LLMs, and ML ethics.

See the Beginner roadmap โ†’

๐Ÿง  CogSci

You are here for Bayesian cognitive science: probabilistic models as theories of minds โ€” generalization, memory, sampling hypotheses, goal inference, and theory of mind.

See the CogSci roadmap โ†’

What you’ll be able to do

By the end of your track you will be able to:

  1. Reason with probability โ€” events, conditioning, and Bayes’ rule, built from counting
  2. Write generative models in code โ€” simulate, trace, and condition with GenJAX
  3. Learn from data the Bayesian way โ€” priors, posteriors, conjugacy, mixtures, and hierarchies
  4. Model structure and dynamics โ€” Bayes nets, causality, Markov chains, and sampling algorithms
  5. Model decisions and agents โ€” utilities, MDPs, reinforcement learning, and inverse inference about goals
  6. (Beginner) Understand modern ML โ€” neural networks, transformers, and what LLMs have to do with Bayes ยท (CogSci) Build computational theories of minds and read the current literature

How the book is organized

Chapters are grouped into Parts (the numbers in the sidebar). Each Part is one arc of Chibany’s year with its own driving question; each Part’s opening page has a small roadmap. Read a Part in order unless your track says otherwise โ€” chapters assume their Part’s earlier episodes.

If you are following the Human and Machine Learning course, each chapter’s closing box links the matching week’s lecture deck, and the in-chapter self-check quizzes are the same polls used live in lecture.

This project is generously funded by the Japanese Probabilistic Computing Consortium Association (JPCCA).