Multi-objective ACO for MOTSP - comparison of best-preforming configurations

The dataset contains results from solving various Multi-Objective Travelling Salesman Problem instances repeatedly by a configurable Ant Colony Optimization algorithm. The algorithm includes a novel modification - a multi-dimensional pheromone. Using irace package (https://github.com/MLopez-Ibanez/irace) we have identified top-performing configurations of the algorithm and compared them with each other and with the configurations initially found by grid search of the hyper-parameter space. The CSV file contains the following information: the configuration used, the instance solved, and the values of the hypervolume-based indicator for the final global Pareto in each of the 20 runs.

本数据集收录了通过可配置蚁群优化（Ant Colony Optimization）算法反复求解各类多目标旅行商问题（Multi-Objective Travelling Salesman Problem）实例所得到的实验结果。该算法包含一项创新性改进——多维信息素。我们借助irace工具包（https://github.com/MLopez-Ibanez/irace）确定了该算法性能最优的配置方案，并将这些方案相互对比，同时与通过超参数空间网格搜索初始得到的配置方案进行比对。本次配套的CSV文件包含以下信息：所采用的算法配置、待求解的问题实例，以及20次独立运行中每次得到的最终全局帕累托（Pareto）解集对应的基于超体积的指标数值。