Combining Reinforcement Learning with Lin-Kernighan-Helsgaun Algorithm for the Traveling Salesman Problem
This work aims to improve the performance of a well-known TSP solver for researchers and practitioners working on combinatorial optimization, representing an incremental improvement to an existing algorithm.
This paper tackles the Traveling Salesman Problem (TSP) by proposing VSR-LKH, a variable strategy reinforced approach that integrates Q-learning, Sarsa, and Monte Carlo reinforcement learning methods with the Lin-Kernighan-Helsgaun (LKH) algorithm. The method replaces LKH's fixed traversal with learned choices, achieving excellent performance on 111 TSPLIB benchmarks with up to 85,900 cities.
We address the Traveling Salesman Problem (TSP), a famous NP-hard combinatorial optimization problem. And we propose a variable strategy reinforced approach, denoted as VSR-LKH, which combines three reinforcement learning methods (Q-learning, Sarsa and Monte Carlo) with the well-known TSP algorithm, called Lin-Kernighan-Helsgaun (LKH). VSR-LKH replaces the inflexible traversal operation in LKH, and lets the program learn to make choice at each search step by reinforcement learning. Experimental results on 111 TSP benchmarks from the TSPLIB with up to 85,900 cities demonstrate the excellent performance of the proposed method.