<ul data-eligibleForWebStory="true"><li>Combinatorial optimization problems are difficult due to their discrete structure and large solution space.</li><li>Deep reinforcement learning (DRL) has shown the ability to learn heuristics from data but can struggle with limited exploration and local optima.</li><li>Genetic Algorithms (GAs) excel in global exploration but are sample inefficient and computationally intensive.</li><li>A new framework called Evolutionary Augmentation Mechanism (EAM) combines DRL efficiency with GA's global search power by refining solutions through genetic operations.</li><li>EAM enhances exploration and speeds up convergence by integrating evolved solutions back into the policy training loop.</li><li>Theoretical analysis ensures stable policy updates by establishing an upper bound on the KL divergence between evolved and policy distributions.</li><li>EAM is versatile and can be used with various DRL solvers like Attention Model, POMO, and SymNCO.</li><li>Extensive testing on benchmark problems like TSP, CVRP, PCTSP, and OP shows EAM improves solution quality and training efficiency compared to baselines.</li></ul>

Synergizing Reinforcement Learning and Genetic Algorithms for Neural Combinatorial Optimization

Discover more