太阳成集团tyc4633(中国)有限公司-百度百科

您当前位置: 首页  >  科学研究  >  学术动态  >  正文

学术动态

计算机科学技术专家讲座(十)—— Ramiro Varela

发布日期:2024-06-17 发布人: 点击量:

报告题目:Evolving Scheduling Heuristics by Means of Hyper-heuristics

报告时间:2024619 900

报告地点:吉林大学中心校区王湘浩楼A521

报告人:Ramiro Varela教授 西班牙奥维耶多大学

报告人简介:

Ramiro Varela is professor of Artificial Intelligence and Computer Science in the Department of Computing of the University of Oviedo. He is member of the intelligent Scheduling and Optimization team (iScOp) and his current research interest includes heuristic search, metaheuristics and hyper-heuristics with application to scheduling and other combinatorial optimization problems. He supervised 5 doctoral thesis and has participated as researcher or responsible in various Spanish national projects and published a number of journal and conference papers on these topics. He is currently coordinator of the doctoral program in computing at the University of Oviedo and Associate Editor of Applied Soft Computing journal. He is member of the Spanish Association for Artificial Intelligence (AEPIA) and the Spanish Society for Computer Science (SCIE).


报告内容简介:

Scheduling problems arise everywhere in industrial environments. They are usually NP-hard problems and so they must be solved by means of heuristic methods that produce reasonable solutions by a limited time. For large problem instances, even the most powerful metaheuristics, as for example memetic algorithms, still require prohibitive execution times. In these cases, greedy algorithms guided by scheduling heuristics, also called schedule builders or schedule generation schemes, are often the only viable solution. The scheduling heuristics may be designed at hand by experts in the problem or, preferably, by means of automatic methods that may capture problem features that are not obvious to the human eye. In this talk we review a Genetic Programming approach developed in the iScOp team to evolve scheduling heuristics in the form of priority rules that allow the greedy algorithm to be adapted to families of problem instances having similar features.


主办单位:太阳成集团tyc4633(中国)有限公司-百度百科

吉林大学软件学院

吉林大学计算机科学技术研究所

符号计算与知识工程教育部重点实验室

仿真技术教育部重点实验室

网络技术及应用软件教育部工程研究中心

     吉林大学国家级计算机实验教学示范中心