← Research
📡

Restless Multi-armed Bandits

Restless multi-armed bandits (RMABs) are a widely used framework for sequential decision-making, with applications in many fields. Their solution, however, is hindered by the exponentially growing state space and the combinatorial action space. As a result, designing efficient planning and learning algorithms for RMABs remains a long-standing challenge.

Convergence-guaranteed decentralized optimization over gossip communication networks with adversarial link failures.

Related Publications

2026
GINO-Q: Learning an asymptotically optimal index policy for restless multi-armed bandits
Gongpu Chen, Soung Chang Liew, and Deniz Gündüz
AAAI 2026 · Singapore
Oral Presentation Reinforcement Learning Restless Multi-armed Bandits
2023
An index policy for minimizing the uncertainty-of-information of Markov sources
Gongpu Chen and Soung Chang Liew
IEEE Transactions on Information Theory, vol. 70, no. 1, pp. 698-721
Communication Network Information Freshness Restless Multi-armed Bandits
2022
Uncertainty-of-Information Scheduling: A Restless Multiarmed Bandit Framework
Gongpu Chen, Soung Chang Liew, and Yulin Shao
IEEE Transactions on Information Theory, vol. 68, no. 9, pp. 6151-6173
Communication Network Information Freshness Restless Multi-armed Bandits