Reactive Sample Size for Heuristic Search in Simulation-based Optimization
This work addresses efficiency challenges in simulation-based optimization for domains like revenue management, though it appears incremental as it builds on existing parametric tests and indifference-zone selection methods.
The paper tackles the problem of determining optimal sample sizes for comparing parameter settings in stochastic simulation-based optimization, presenting a reactive algorithm that improves efficiency and robustness in experiments with benchmark functions and hotel revenue management simulations.
In simulation-based optimization, the optimal setting of the input parameters of the objective function can be determined by heuristic optimization techniques. However, when simulators model the stochasticity of real-world problems, their output is a random variable and multiple evaluations of the objective function are necessary to properly compare the expected performance of different parameter settings. This paper presents a novel reactive sample size algorithm based on parametric tests and indifference-zone selection, which can be used for improving the efficiency and robustness of heuristic optimization methods. The algorithm reactively decides, in an online manner, the sample size to be used for each comparison during the optimization according to observed statistical evidence. Tests employ benchmark functions extended with artificial levels of noise and a simulation-based optimization tool for hotel revenue management. Experimental results show that the reactive method can improve the efficiency and robustness of simulation-based optimization techniques.