Enhancing Novel Clean Room Learning Metaheuristic Algorithm on Noisy Response Surfaces: Parameter Design through Dual Response Optimization

Abstract

Metaheuristics, such as the clean room learning approach, are utilized in order to tackle complex optimization issues. The potential hindrance of performance might arise from the existence of response surfaces that exhibit noise, which is manifested as random disturbances introduced during the evaluation of objective functions. The objective of this work is to determine algorithm parameters by employing dual response optimization approaches. The optimization of dual response entails the concurrent assessment of both the estimated mean difference from the target and the standard deviation. The program adeptly oversees the distribution of resources, skillfully balancing the pursuit of research in uncharted domains and capitalizing on the insights derived from analyzing flawed information. The effectiveness of the method has been shown by experiments done on benchmark functions with different levels of noise. The technique discussed above demonstrates higher performance when compared to fixed-parameter competitors in terms of its capacity to withstand noise, achieve high precision, and maintain stability during the process of searching for optimal solutions. Our study is centered around the examination of the effects of different forms of noise on algorithmic performance, alongside the determination of the most favorable parameter values for different levels of noise. The current research highlights the lack of structure inherent in the algorithm. This paper gives a comprehensive investigation into the optimization of dual responses in the presence of surface noise, with the objective of improving the efficacy of metaheuristic algorithms employed in clean room learning. The technique presented in this study aims to tackle the issue of noise in the objective function, which ultimately results in improved optimization results. The findings of this research study provide a significant contribution to the advancement of metaheuristic optimization approaches and has extensive applicability.