Motivation 

The aim of this work is to utilize machine learning techniques to enhance the efficiency and effectiveness of primal heuristics for solving challenging mixed-integer linear programming problems. Here, we use ML techniques to optimize the configuration parameters of the heuristics, thereby achieving better performance and results.

ML4CO Competition

primal task:

primal heuristics at the root node

Final Results of Primal Task

Three datasets:

• item placement
• load balancing
• anonymous

Summary

Team members

Linxin Yang, Sha Lai, Akang Wang, Xiaodong Luo from SRIBD；

Xiang Zhou, Haohan Huang, Shengcheng Shao, Yuanming Zhu, Dong Zhang from Huawei.

Award

We won the first place in the NeurIPS 2021 ML4CO(Machine Learning for Combinatorial Optimization) Competition.