Online Control of Adaptive Large Neighborhood Search using Deep Reinforcement Learning to Enhance Combinatorial Optimization

The proposed DR-ALNS method leverages Deep Reinforcement Learning to dynamically select operators, adjust destroy severity, and control the acceptance criterion within the Adaptive Large Neighborhood Search (ALNS) algorithm, leading to more effective solutions for combinatorial optimization problems.