Improving the Fine-Tuning of Metaheuristics: An Approach Combining Design of Experiments and Racing Algorithms

Abstract

Usually, metaheuristic algorithms are adapted to a large set of problems by applying few modifications on parameters for each specific case. However, this flexibility demands a huge effort to correctly tune such parameters. Therefore, the tuning of metaheuristics arises as one of the most important challenges in the context of research of these algorithms. Thus, this paper aims to present a methodology combining Statistical and Artificial Intelligence methods in the fine-tuning of metaheuristics. The key idea is a heuristic method, called Heuristic Oriented Racing Algorithm (HORA), which explores a search space of parameters looking for candidate configurations close to a promising alternative. To confirm the validity of this approach, we present a case study for finetuning two distinct metaheuristics: Simulated Annealing (SA) and Genetic Algorithm (GA), in order to solve the classical traveling salesman problem. The results are compared considering the same metaheuristics tuned through a racing method. Broadly, the proposed approach proved to be effective in terms of the overall time of the tuning process. Our results reveal that metaheuristics tuned by means of HORA achieve, with much less computational effort, similar results compared to the case when they are tuned by the other fine-tuning approach.

A Survey of Automatic Parameter Tuning Methods for Metaheuristics

“…4) HORA: The so-called heuristic oriented racing algorithm (HORA) introduced by Barbosa and Senne [85], [86] is a relatively new heuristic search method for parameter tuning. HORA is an iterative algorithm which dynamically creates candidate configurations and uses the racing method to evaluate them.…”

“…These identified parameter settings ensure diversity of the parameters and they are used to define the upper and lower bounds of each parameter. After that, HORA enters an iterative procedure consisting of: 1) dynamically creating new candidates in the neighborhood of some best known candidate configurations, i.e., configurations that are preferred in evaluation and 2) evaluating the set of candidate configurations with racing method to discard poor ones according to the statistical evidences [85], [86]. By this repeated procedure, HORA algorithm consistently finds better candidate configurations.…”

“…In [85] and [86] a case study about HORA, in which a bruteforce and racing methods were also considered for comparison, was performed on GA and SA with instances selected from the OR-Library [87]. The results of the case study showed that HORA achieve similar (or even better) results but with much lower computational cost compared to other approaches.…”

An evaluation of three DoE-guided meta-heuristic-based solution methods for a three-echelon sustainable distribution network

“…One of the major challenges in designing an evolutionary-based meta-heuristic solution method is the details of its design, in particular parameter setting (Hoos 2011; Barbosa and Senne 2017). A range of methods from more conventional approaches with low-mutation rates to uniform crossover choices and experimental designs have been introduced and adopted in practice (Eiben and Smit 2011).…”

Reliability aware multi-objective predictive control for wind farm based on machine learning and heuristic optimizations

“…Based on the results and by conducting the design of experiments repeatedly, the suitable parameters can be identified. The interested readers can refer to [46] for more details about more fine tuning of the parameters.…”