Title: Enhancing Constraint Programming via Supervised Learning for Job Shop Scheduling

Abstract: Constraint programming (CP) is an effective technique for solving constraint satisfaction and optimization problems. CP solvers typically use a variable ordering strategy to select which variable to explore first in the solving process, which has a large impact on the efficacy of the solvers. In this paper, we propose a novel variable ordering strategy based on supervised learning to solve job shop scheduling problems. We develop a classification model and a regression model to predict the optimal solution of a problem instance, and use the predicted solution to order variables for CP solvers. We show that training machine learning models is very efficient and can achieve a high accuracy. Our extensive experiments demonstrate that the learned variable ordering methods perform competitively compared to four existing methods. Finally, we show that hybridising the machine learning-based variable ordering methods with traditional domain-based methods is beneficial.