Applying machine learning techniques to metaheuristic algorithms to solve combinatorial optimization problems

• Main Authors: Garcia - Conejeros, Jose
• Other Authors: Pontificia Universidad Catolica De Valparaiso
• Format: Report
• Language: unknown
• Published: 2023
• Subjects: Machine Learning Metaheuristics Combinatorial problems; Ingenieria De Sistemas Y Comunicaciones; Magallanes; Valparaíso; Bío Bío; General Bernardo O'Higgins; Antártica
• Online Access: https://hdl.handle.net/10533/48336

3 The machine learning binary operator allows improving the quality of the solutions or the convergence time of the metaheuristic algorithms when solving COP. TOTAL In the different experiments carried out, the Multidimensional Knapsack, Multi-demand multidimensional knapsack, Set-union knapsack, and set covering problems obtained improvements in the quality of the results when they were compared against different operators such as random operators, function transfers, or against algorithms of the state-of-the-art. 4 To use K-nearest neighbors algorithm to create a local perturbation operator that allows improving the quality of the solutions or the convergence time of the metaheuristic algorithms when they solve COP. TOTAL The KNN technique was used as a perturbation operator in the multidimensional knapsack and set covering problems, obtaining satisfactory results. Particularly, the results obtained from the knapsack were published in the article entitled "Enhancing a machine learning binarization framework by perturbation operators: analysis on the multidimensional knapsack problem" and in the case of set covering it was published with the title "An analysis of a KNN perturbation operator: an application to the binarization of continuous metaheuristics" 5 To apply the hybrid metaheuristics to knapsack, traveling salesman and Transit route network design problems. TOTAL In the benchmark problems as mentioned above, it was successfully applied to: 1. Multidimensional knapsack problem 2. Set covering problem 3. Set union knapsack problem 4. Multi-demand multidimensional knapsack problem. In the case of real-life problems, it was applied to obtain satisfactory results: 1. Crew scheduling problems 2. Multi-Objective buttressed wall problems (saved as a single objective) 3. Multi-objective steel-concrete composite bridges (saved as a single objective) Otro(s) aspecto(s) que Ud. considere importante(s) en la evaluación del cumplimiento de objetivos planteados en la propuesta original o en las modificaciones ...