Learning to Perform Local Rewriting for Combinatorial Optimization

TL;DR

NeuRewriter is a neural network-based reinforcement learning method that learns to iteratively improve solutions for combinatorial optimization problems by selecting heuristics and rewriting local solution components.

Contribution

It introduces a neural policy that learns to perform local rewrites in combinatorial problems, outperforming existing heuristics and neural baselines across multiple tasks.

Findings

NeuRewriter outperforms Z3 in expression simplification.

NeuRewriter surpasses DeepRM and OR-tools in online job scheduling.

NeuRewriter exceeds recent neural baselines and OR-tools in vehicle routing.

Abstract

Search-based methods for hard combinatorial optimization are often guided by heuristics. Tuning heuristics in various conditions and situations is often time-consuming. In this paper, we propose NeuRewriter that learns a policy to pick heuristics and rewrite the local components of the current solution to iteratively improve it until convergence. The policy factorizes into a region-picking and a rule-picking component, each parameterized by a neural network trained with actor-critic methods in reinforcement learning. NeuRewriter captures the general structure of combinatorial problems and shows strong performance in three versatile tasks: expression simplification, online job scheduling and vehicle routing problems. NeuRewriter outperforms the expression simplification component in Z3; outperforms DeepRM and Google OR-tools in online job scheduling; and outperforms recent neural baselines and Google OR-tools in vehicle routing problems.