Hyper heuristic based on great deluge and its variants for exam timetabling problem
This work addresses the time-consuming and hard university timetabling problem, but it is incremental as it applies existing variants to a known domain.
The paper tackled the exam timetabling problem by developing hyper-heuristic methods using Great Deluge and its variants as move acceptance strategies combined with reinforcement learning, reporting best results and comparisons on benchmark datasets.
Today, University Timetabling problems are occurred annually and they are often hard and time consuming to solve. This paper describes Hyper Heuristics (HH) method based on Great Deluge (GD) and its variants for solving large, highly constrained timetabling problems from different domains. Generally, in hyper heuristic framework, there are two main stages: heuristic selection and move acceptance. This paper emphasizes on the latter stage to develop Hyper Heuristic (HH) framework. The main contribution of this paper is that Great Deluge (GD) and its variants: Flex Deluge(FD), Non-linear(NLGD), Extended Great Deluge(EGD) are used as move acceptance method in HH by combining Reinforcement learning (RL).These HH methods are tested on exam benchmark timetabling problem and best results and comparison analysis are reported.