Simulation-guided Beam Search for Neural Combinatorial Optimization

TL;DR

This paper introduces simulation-guided beam search (SGBS), a novel method that combines neural policies with simulation to improve combinatorial optimization, and enhances it with active search for better solutions.

Contribution

It proposes SGBS, a new search algorithm for neural combinatorial optimization that integrates simulation and beam search, and hybridizes it with active search for improved performance.

Findings

SGBS significantly improves solution quality on benchmark problems.

Hybrid SGBS and EAS outperform existing neural and heuristic methods.

The approach achieves better solutions within reasonable runtime constraints.

Abstract

Neural approaches for combinatorial optimization (CO) equip a learning mechanism to discover powerful heuristics for solving complex real-world problems. While neural approaches capable of high-quality solutions in a single shot are emerging, state-of-the-art approaches are often unable to take full advantage of the solving time available to them. In contrast, hand-crafted heuristics perform highly effective search well and exploit the computation time given to them, but contain heuristics that are difficult to adapt to a dataset being solved. With the goal of providing a powerful search procedure to neural CO approaches, we propose simulation-guided beam search (SGBS), which examines candidate solutions within a fixed-width tree search that both a neural net-learned policy and a simulation (rollout) identify as promising. We further hybridize SGBS with efficient active search (EAS), where SGBS enhances the quality of solutions backpropagated in EAS, and EAS improves the quality of the policy used in SGBS. We evaluate our methods on well-known CO benchmarks and show that SGBS significantly improves the quality of the solutions found under reasonable runtime assumptions.