Derivative-Free & Order-Robust Optimisation
This addresses a rarely considered challenge in optimization for scenarios where order matters, offering a new approach for robust decision-making.
The paper tackles the problem of order-robust optimization in adversarial environments by formalizing it as online learning to minimize simple regret, and introduces Vroom, a derivative-free algorithm that achieves vanishing regret in non-stationary settings and favorable rates under stochastic processes.
In this paper, we formalise order-robust optimisation as an instance of online learning minimising simple regret, and propose Vroom, a zero'th order optimisation algorithm capable of achieving vanishing regret in non-stationary environments, while recovering favorable rates under stochastic reward-generating processes. Our results are the first to target simple regret definitions in adversarial scenarios unveiling a challenge that has been rarely considered in prior work.