About This Book
Optimal Control Using Causal Agents is a translation manual between Causal Inference and Reinforcement Learning, written for researchers and practitioners already familiar with either Causal Inference or Reinforcement Learning. Each theoretical connection is grounded in concrete applications ranging from clinical decision-making and Brazilian Jiu-Jitsu strategy to GARCH financial modeling. The book uses R and Python for implementation.
Optimal Control Using Causal Agents is a rosetta stone, not a comprehensive textbook (see the Causal AI book developed by Bareinboim, et. al. for that). In short, Optimal Control Using Causal Agents builds bridges between existing concepts instead of constructing concepts from the ground up. The goal is for causal inference practitioners to gain instant access to reinforcement learning's computational tools by seeing how these tools share ideas with their causal methods, and vice versa. The result is a practical guide for navigating 70 years of parallel mathematical development that has been artificially separated by academic boundaries. Perhaps the most important contribution is a comprehensive translation table which creates a map between notational conventions in both fields.
Publisher: CRC Press
Expected Publication: Spring 2026
Series: Chapman & Hall/CRC Machine Learning & Pattern Recognition
Editor: Lara Spieker
Table of Contents
Part I: The Divide
- Introduction
- Programming
Part II: Foundations: Causal Inference
- Causal Inference With Randomization
- Causal Effects Without Randomization
Part III: Foundation and Empire: Bridges From Reinforcement Learning to Causal Inference
- Tabular Reinforcement Learning
- Reinforcement Learning Models
Part IV: Second Foundation: Synthesis
- Beyond Markovian Dynamics
- Beyond Ignorable Missing Data
Resources
Preliminary chapters and accompanying code implementations will be available shortly.