Integration of Machine Learning into Local Search Meta-heuristics for the Sports Scheduling Problem

Abstract

This thesis addresses the Traveling Tournament Problem (TTP), an NP-hard combinatorial optimization challenge in sports scheduling. The TTP is crucial in the sports community due to its impact on league management budgets, where suboptimal schedules can lead to significant financial losses. To tackle this problem, we propose an innovative approach that integrates machine learning and deep learning techniques to enhance the performance of Stochastic Local Search (SLS) and Biogeography-Based Optimization (BBO) algorithms. It is well known that the main challenge in the evolutionary algorithms is to set the best parameters. we employ a deep learning model for parameter tuning. Additionally, we proposed a novel approach to incorporate reinforcement learning within the SLS algorithm during the exploitation phase to enhance its performance. The computational experiments indicate that our method produces competitive and promising results, showcasing its potential for effective TTP optimization.