Offline-Learning-Based Automated Generation of Metaheuristics

组合优化启发式算法的设计和开发是一个既耗时又昂贵的过程，因为它需要昂贵而难得的专业知识。本研究的目的是创建一个可重用的基于离线学习的元启发式生成器。我们将定义一个可配置的元启发式算法，它能够对广泛的标准元启发式算法进行建模，研究专门用于该元启发式算法的算法配置方法，为用户提供一个方便的界面来应用该框架来解决他们的问题，并使用该框架来客观地比较各种标准元启发式方法的性能。我们将把这个框架应用于著名的组合优化问题，以评估其性能并产生最先进的解决方法。

Design and development of combinatorial optimisation heuristics is a time-consuming and expensive process as it requires expensive and rare expertise. The aim of this research is to create a reusable offline-learning-based metaheuristic generator. We will define a configurable metaheuristic that is capable of modelling a wide range of standard metaheuristics, research algorithm configuration methods specialised for this metaheuristic, provide a convenient interface for the users to apply this framework to their problems and use the framework to objectively compare the performance of various standard metaheuristic approaches. We will apply this framework to well-known combinatorial optimisation problems to evaluate its performance and produce state-of-the-art solution methods.