PoMo

该数据集是为了评估在车辆路径问题（VRP）任务中，包括旅行商问题（TSP）和带容量限制的车辆路径问题（CVRP）等各种神经方法与传统求解器在泛化能力方面的表现。POMO是一种基于强大构建的神经方法，专门用于解决车辆路径问题（VRP），并针对不同规模和分布的泛化能力进行了设计。This dataset is developed to evaluate the generalization performance of various neural methods and traditional solvers when applied to vehicle routing problem (VRP) tasks, including representative subproblems such as the Traveling Salesman Problem (TSP) and Capacitated Vehicle Routing Problem (CVRP). POMO is a robustly constructed neural method specifically designed for solving vehicle routing problems (VRP), and it is tailored to optimize generalization performance across different problem scales and distributions.

搜集汇总

背景与挑战

背景概述

该数据集是名为“POMO”的GitHub代码仓库，实现了NeurIPS 2020论文“Policy Optimization with Multiple Optima for Reinforcement Learning”中的算法。仓库包含两个版本：原始交互式Jupyter Notebook代码和更新后的结构化Python代码，适用于强化学习研究。