A survey for solving mixed integer programming via machine learning

Machine learning (ML) has been recently introduced to solving optimization problems, especially for combinatorial (CO) tasks. In this paper, we survey the trend of leveraging ML solve mixed-integer programming problem (MIP). Theoretically, MIP is an NP-hard problem, and most CO problems can be formulated as MIP. Like other human-designed heuristic algorithms rely on good initial solutions cost a lot computational resources. Therefore, researchers consider applying machine methods since ML-enhanced approaches provide solution based typical patterns from training data. Specifically, first introduce formulation preliminaries representative traditional solvers. Then, show integration with detailed discussions related learning-based methods, which further classified into exact algorithms. Finally, propose outlook solvers, direction toward more beyond MIP, mutual embrace solvers components. We maintain list papers that utilize technologies available at https://github.com/Thinklab-SJTU/awesome-ml4co.