Adaptive Bayesian Optimization Algorithm for Unpredictable Business Environments

Sarit Maitra
2024-01-21
Abstract:This paper presents an innovative optimization framework and algorithm based on the Bayes theorem, featuring adaptive conditioning and jitter. The adaptive conditioning function dynamically modifies the mean objective function in each iteration, enhancing its adaptability. The mean function, representing the model's best estimate of the optimal value for the true objective function, is adjusted based on observed data. The framework also incorporates an adaptive acquisition jitter function, enhancing adaptability by adjusting the jitter of the acquisition function. It also introduces a robust objective function with a penalty term, aiming to generate robust solutions under uncertainty. The evaluation of the framework includes single-objective, decoupled multi-objective, and combined multi-objective functions. Statistical analyses, including t-statistics, p-values, and effect size measures, highlight the superiority of the proposed framework over the original Bayes optimization. The adaptive nature of the conditioning function allows the algorithm to seamlessly incorporate new data, making it particularly beneficial in dynamic optimization scenarios.
Optimization and Control
What problem does this paper attempt to address?