POND: Pessimistic-Optimistic oNline Dispatching

TL;DR

This paper introduces POND, an online dispatching algorithm that balances pessimistic and optimistic strategies to minimize regret and constraint violations in uncertain, real-time decision-making scenarios.

Contribution

The paper presents a novel online dispatching algorithm, POND, with proven optimal regret and constraint violation bounds, applicable to unknown distribution settings.

Findings

POND achieves $O(\sqrt{T})$ regret.

POND maintains $O(1)$ constraint violation.

Experimental results show low regret and minimal violations.

Abstract

This paper considers constrained online dispatching with unknown arrival, reward and constraint distributions. We propose a novel online dispatching algorithm, named POND, standing for Pessimistic-Optimistic oNline Dispatching, which achieves $O(\sqrt{T})$ regret and $O(1)$ constraint violation. Both bounds are sharp. Our experiments on synthetic and real datasets show that POND achieves low regret with minimal constraint violations.