Inférence Causale pour l'Évaluation des Politiques pour Ingénieurs en Apprentissage par Renforcement Débutants
Master causal inference techniques to accurately evaluate Reinforcement Learning policies from observational data, moving beyond correlation to understand true policy impact and inform robust system design.
...
Share
Foundations of Causal Inference for RL Policy Evaluation
Unit 1: Introduction to Causal Thinking
Correlation vs. Causation
The Causal Question
Unit 2: The Counterfactual Problem
What If? The Counterfactual
The Fundamental Problem
Unit 3: Causal Graphs: DAGs
Drawing Causal Maps
Confounding Explained
Paths to Bias
DAGs for RL Policies
Applied Causal Methods for RL Policy Evaluation
Unit 1: Introduction to Off-Policy Evaluation
Why Off-Policy Eval?
The OPE Problem
Unit 2: Inverse Probability Weighting (IPW)
IPW: The Basics
IPW for RL Policies
IPW: Practical Aspects
Unit 3: Doubly Robust (DR) Estimation
DR: Combining Strengths
DR for RL Policies
Interpreting OPE Results