DeepCO: Offline Combinatorial Optimization Framework Utilizing Deep Learning

TL;DR

DeepCO is a novel offline combinatorial optimization framework that leverages deep learning to improve solutions in complex industrial problems, demonstrated through a TSP variant reducing route length by 5.7%.

Contribution

The paper introduces DeepCO, a new offline deep learning-based framework for combinatorial optimization, and designs an offline TSP variant for evaluation.

Findings

Outperforms baseline methods in offline TSP with 5.7% route length reduction

Demonstrates potential for real-world industrial applications

Effective with limited historical data

Abstract

Combinatorial optimization serves as an essential part in many modern industrial applications. A great number of the problems are offline setting due to safety and/or cost issues. While simulation-based approaches appear difficult to realise for complicated systems, in this research, we propose DeepCO, an offline combinatorial optimization framework utilizing deep learning. We also design an offline variation of Travelling Salesman Problem (TSP) to model warehouse operation sequence optimization problem for evaluation. With only limited historical data, novel proposed distribution regularized optimization method outperforms existing baseline method in offline TSP experiment reducing route length by 5.7% averagely and shows great potential in real world problems.