Multi-Agent Reinforcement Learning and Bandit Learning

Program

Learning and Games

Location

Calvin Lab Auditorium

Date

Monday, May 2 – Thursday, May 5, 2022

Back to calendar

8:30 – 9 a.m.

Coffee and Check-In

9 – 9:05 a.m.

Opening Remarks

9:05 – 9:45 a.m.

"Calibeating": Beating Forecasters at Their Own Game

Sergiu Hart (Hebrew University of Jerusalem)

9:45 – 10:15 a.m.

Break

10:15 – 11 a.m.

V-Learning: Simple, Efficient, Decentralized Algorithm for Multiagent RL

Chi Jin (Princeton University)

11:30 a.m. – 12:15 p.m.

What is the Statistical Complexity of Reinforcement Learning?

Sham Kakade (Harvard and MSR)

12:15 – 2 p.m.

Lunch

2 – 2:30 p.m.

No-Regret Learning in Time-Varying Zero-Sum Games

Haipeng Luo (University of Southern California)

2:30 – 3 p.m.

Policy Gradients in General-Sum Dynamic Games: When Do They Even Converge?

Eric Mazumdar (Caltech)

3 – 3:15 p.m.

Break

3:15 – 4 p.m.

The Complexity of Infinite-Horizon General-Sum Stochastic Games: Turn-Based and Simultaneous Play

Vidya Muthukumar (Georgia Institute of Technology)

4 – 5 p.m.

Reception

11 – 11:30 a.m.

Break

8:30 – 9 a.m.

Coffee and Check-In

9 – 9:45 a.m.

The Complexity of Markov Equilibrium in Stochastic Games

Noah Golowich (MIT)

9:45 – 10:15 a.m.

Break

10:15 – 11 a.m.

A Regret Minimization Approach to Mutli-Agent Control and RL

Elad Hazan (Princeton University and Google Research)

11 – 11:30 a.m.

Break

11:30 a.m. – 12:15 p.m.

Multi-Agent Reinforcement Learning in the High Population Regime

Tamer Başar (University of Illinois Urbana-Champaign)

12:15 – 2 p.m.

Lunch

2 – 2:30 p.m.

Learning Automata as Building Blocks for MARL

Ann Nowe (Vrije Universiteit Brussel)

2:30 – 3 p.m.

On Rewards in Multi-Agent Systems

Doina Precup (McGill University / DeepMind Montreal)

3 – 3:30 p.m.

Break

3:30 – 4 p.m.

Independent Learning in Stochastic Games

Kaiqing Zhang (MIT)

4 – 4:30 p.m.

Sequential Information Design: Markov Persuasion Process and Its Efficient Reinforcement Learning

Zhuoran Yang (UC Berkeley)

4:30 – 5:30 p.m.

Open Problem Session

8:30 – 9 a.m.

Coffee and Check-In

9 – 9:45 a.m.

Multi-Agent Reinforcement Learning Towards Zero-Shot Communication

Kalesha Bullard (DeepMind)

9:45 – 10:15 a.m.

Break

10:15 – 11 a.m.

Kernelized Multiplicative Weights for 0/1-Polyhedral Games: Bridging the Gap Between Learning in Extensive-Form and Normal-Form Games

Gabriele Farina (Carnegie Mellon University)

11 – 11:30 a.m.

Break

11:30 a.m. – 12 p.m.

General Game-Theoretic Multiagent Reinforcement Learning

Marc Lanctot (DeepMind)

12 – 2 p.m.

Lunch

2 – 2:30 p.m.

When Is Offline Two-Player Zero-Sum Markov Game Solvable?

Simon Du (University of Washington)

2:30 – 3 p.m.

Variants and Invariants in No-Regret Algorithms

Sylvain Sorin (Sorbonne Universite)

3 – 3:30 p.m.

Break

3:30 – 4 p.m.

What Does Machine Learning Offer Game Theory (And Vice Versa)?

Alex Peysakhovich (Meta AI)

4 – 4:30 p.m.

Multi-Player Bandits With No Collisions

Mark Sellke (Stanford)

8:30 – 9 a.m.

Coffee and Check-In

9 – 9:45 a.m.

Global Convergence of Multi-Agent Policy Gradient in Markov Potential Games

Ioannis Panageas (UC Irvine)

9:45 – 10:15 a.m.

Break

10:15 – 11 a.m.

Learning and Equilibrium Refinements

Drew Fudenberg (MIT)

11 – 11:30 a.m.

Break

11:30 a.m. – 12:15 p.m.

No-Regret Learning in Extensive-Form Games

Amy Greenwald (Brown University)

12:15 – 2 p.m.

Lunch

2 – 2:30 p.m.

Learning Decentralized Policies in Multiagent Systems: How to Learn Efficiently and What Are the Learned Policies?

Na Li (Harvard University)

2:30 – 2:45 p.m.

Break

2:45 – 3:30 p.m.

The Role of Conventions in Adaptive Human-AI Interaction

Dorsa Sadigh (Stanford University)