Learning Combinatorial Optimization Algorithms over Graphs

TL;DR

This paper introduces a reinforcement learning and graph embedding framework to automatically learn heuristics for solving recurring combinatorial optimization problems on graphs, replacing manual heuristic design.

Contribution

It presents a novel combination of reinforcement learning and graph embeddings to learn effective algorithms for various graph-based optimization problems.

Findings

Successfully applied to Minimum Vertex Cover, Maximum Cut, and Traveling Salesman problems.

Learned policies outperform traditional heuristics in several cases.

Framework generalizes across different combinatorial problems.

Abstract

The design of good heuristics or approximation algorithms for NP-hard combinatorial optimization problems often requires significant specialized knowledge and trial-and-error. Can we automate this challenging, tedious process, and learn the algorithms instead? In many real-world applications, it is typically the case that the same optimization problem is solved again and again on a regular basis, maintaining the same problem structure but differing in the data. This provides an opportunity for learning heuristic algorithms that exploit the structure of such recurring problems. In this paper, we propose a unique combination of reinforcement learning and graph embedding to address this challenge. The learned greedy policy behaves like a meta-algorithm that incrementally constructs a solution, and the action is determined by the output of a graph embedding network capturing the current state of the solution. We show that our framework can be applied to a diverse range of optimization problems over graphs, and learns effective algorithms for the Minimum Vertex Cover, Maximum Cut and Traveling Salesman problems.