Reinforcement Learning Enhanced Heuristic Search for CombinatorialOptimizationReinforcement Learning Enhanced Heuristic Search for CombinatorialOptimization (Reinforcement Learning gebaseerde heuristieken voorcombinatorische optimalisatie) KU Leuven
The present thesis describes the use of reinforcement learning to enhance heuristic search for solving complex (real-world) optimization problems such as (project) scheduling, routing and assignment. Heuristic search methods are known to deliver good results in a reasonable amount of calculation time, without any guarantee of optimality. Often they require careful parameter tuning to obtain good results. Reinforcement learning methods on the ...