Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

SORREL: Suboptimal-Demonstration-Guided Reinforcement Learning for Learning to Branch

About

Mixed Integer Linear Program (MILP) solvers are mostly built upon a Branch-and-Bound (B\&B) algorithm, where the efficiency of traditional solvers heavily depends on hand-crafted heuristics for branching. The past few years have witnessed the increasing popularity of data-driven approaches to automatically learn these heuristics. However, the success of these methods is highly dependent on the availability of high-quality demonstrations, which requires either the development of near-optimal heuristics or a time-consuming sampling process. This paper averts this challenge by proposing Suboptimal-Demonstration-Guided Reinforcement Learning (SORREL) for learning to branch. SORREL selectively learns from suboptimal demonstrations based on value estimation. It utilizes suboptimal demonstrations through both offline reinforcement learning on the demonstrations generated by suboptimal heuristics and self-imitation learning on past good experiences sampled by itself. Our experiments demonstrate its advanced performance in both branching quality and training efficiency over previous methods for various MILPs.

Shengyu Feng, Yiming Yang• 2024

Related benchmarks

TaskDatasetResultRank
Mixed Integer Linear Programming SolvingSet Covering 1000x500 (Easy)
Search Nodes174.3
13
Integer Programming BranchingCauctions Hard
Time204.3
9
Integer Programming BranchingIndset Medium
Time100.3
9
Integer Programming BranchingIndset (Hard)
Time2.13e+3
9
Integer Programming BranchingFacilities Easy
Time30.96
9
Integer Programming BranchingCauctions Medium
Time15.62
9
Integer Programming BranchingFacilities Medium
Time (s)124.7
9
Integer Programming BranchingSetcover Hard
Time1.17e+3
9
Integer Programming BranchingCauctions Easy
Time1.27
9
Integer Programming BranchingIndset (Easy)
Time16.63
9
Showing 10 of 12 rows

Other info

Follow for update