Adversarial Instance Generation and Robust Training for Neural Combinatorial Optimization with Multiple Objectives

TL;DR

A preference-based adversarial attack to generate hard instances that expose solver weaknesses, and a defense strategy that integrates hardness-aware preference selection into adversarial training to reduce overfitting to restricted preference regions and improve out-of-distribution performance.

Abstract

Deep reinforcement learning (DRL) has shown great promise in addressing multi-objective combinatorial optimization problems (MOCOPs). Nevertheless, the robustness of these learning-based solvers has remained insufficiently explored, especially across diverse and complex problem distributions. In this paper, we propose a unified robustness-oriented framework for preference-conditioned DRL solvers for MOCOPs. Within this framework, we develop a preference-based adversarial attack to generate hard instances that expose solver weaknesses, and quantify the attack impact by the resulting degradation on Pareto-front quality. We further introduce a defense strategy that integrates hardness-aware preference selection into adversarial training to reduce overfitting to restricted preference regions and improve out-of-distribution performance. The experimental results on multi-objective traveling salesman problem (MOTSP), multi-objective capacitated vehicle routing problem (MOCVRP), and multi-objective knapsack problem (MOKP) verify that our attack method successfully learns hard instances for different solvers. Furthermore, our defense method significantly strengthens the robustness and generalizability of neural solvers, delivering superior performance on hard or out-of-distribution instances.

Figures (5)

• Figure 1: Results for Varying $c_{\text{DIST}}$ in Gaussian Mixture Generator.
• Figure 2: Attack and Defense of Neural Solvers for MOCOP.
• Figure 3: Impact of Iteration Counts on HV and Gap.
• Figure 4: HV Gaps for Different $\alpha$.
• Figure 5: Benchmark Performance Comparison on HV Metric.