Simons Institute Wiki

After brainstorming about the topics listed below, we decided to run two regular reading groups.

1. Deep RL and Function Approximation: every Wednesday, 9 am to 11 am Pacific Time
2. Causality and Econometrics: every Friday, 9 am to 11 am Pacific Time

Core RL

• Online exploration
• batch RL
• "single sample path"
• infinite horizon
• beyond MDPs (POMDPs and more)
• Pure exploration
• function approximation in RL
• Distributional RL
• Temporal credit assignment
• LP approaches to RL
• causality and RL
• Hierarchical RL

Control and Dynamical Systems

• Unbounded cost and stability
• system identification/LQR
• Nonlinear control

Societal Challenges

• RL and law
• RL and society
• Robustness and safety in RL

Game Theory and RL

• multi-agent control and RL (cooperative and/or competitive)

Statistics

• Max of means (and other non-regular estimation problems)