Memory Length in Hyper-heuristics: An Empirical Study

Hyper-heuristics are an emergent optimisation methodology which aims to give a higher level of flexibility and domain-independence than is currently possible. Hyper-heuristics are able to adapt to the different problems or problem instances by dynamically choosing between heuristics during the search. This paper is concerned with the issues of memory length on the performance of hyper-heuristics. We focus on a recently proposed simulated annealing hyper-heuristic and choose a set of hard university course timetabling problems as the test bed for this empirical study. The experimental results show that the memory length can affect the performance of hyper-heuristics and a good choice of memory length is able to improve solution quality. Finally, two dynamic approaches are investigated and one of the approaches is shown to be able to produce promising results without introducing extra sensitive algorithmic parameters.

[1]  Michael Sampels,et al.  A MAX-MIN Ant System for the University Course Timetabling Problem , 2002, Ant Algorithms.

[2]  Raymond S. K. Kwan,et al.  Distributed Choice Function Hyper-heuristics for Timetabling and Scheduling , 2004, PATAT.

[3]  Graham Kendall,et al.  Search Methodologies: Introductory Tutorials in Optimization and Decision Support Techniques , 2013 .

[4]  Philipp Kostuch,et al.  The University Course Timetabling Problem with a 3-phase approach , 2007 .

[5]  Peter Ross,et al.  Solving a Real-World Problem Using an Evolving Heuristically Driven Schedule Builder , 1998, Evolutionary Computation.

[6]  Richard S. Sutton,et al.  Reinforcement Learning: An Introduction , 1998, IEEE Trans. Neural Networks.

[7]  Thomas G. Dietterich,et al.  High-Performance Job-Shop Scheduling With A Time-Delay TD(λ) Network , 1995, NIPS 1995.

[8]  Alexander Nareyek,et al.  Choosing search heuristics by non-stationary reinforcement learning , 2004 .

[9]  Graham Kendall,et al.  A Tabu-Search Hyperheuristic for Timetabling and Rostering , 2003, J. Heuristics.

[10]  Dushyant Sharma,et al.  A composite very large-scale neighborhood structure for the capacitated minimum spanning tree problem , 2003, Oper. Res. Lett..

[11]  J. Christopher Beck,et al.  APPLYING MACHINE LEARNING TO LOW‐KNOWLEDGE CONTROL OF OPTIMIZATION ALGORITHMS , 2005, Comput. Intell..

[12]  Barry McCollum,et al.  University Timetabling: Bridging the Gap between Research and Practice , 2006 .

[13]  Mauro Birattari,et al.  An effective hybrid algorithm for university course timetabling , 2006, J. Sched..

[14]  Pierre Hansen,et al.  Variable Neighborhood Search , 2018, Handbook of Heuristics.

[15]  Peter Ross,et al.  Learning a Procedure That Can Solve Hard Bin-Packing Problems: A New GA-Based Approach to Hyper-heuristics , 2003, GECCO.

[16]  Wei Zhang,et al.  A Reinforcement Learning Approach to job-shop Scheduling , 1995, IJCAI.

[17]  Xin Yao,et al.  Stochastic ranking for constrained evolutionary optimization , 2000, IEEE Trans. Evol. Comput..

[18]  Sanja Petrovic,et al.  Case-based heuristic selection for timetabling problems , 2006, J. Sched..

[19]  Nysret Musliu,et al.  Local search for shift design , 2004, Eur. J. Oper. Res..

[20]  Graham Kendall,et al.  Hyper-Heuristics: An Emerging Direction in Modern Search Technology , 2003, Handbook of Metaheuristics.

[21]  Edmund K. Burke,et al.  A simulated annealing based hyperheuristic for determining shipper sizes for storage and transportation , 2007, Eur. J. Oper. Res..

[22]  Celso C. Ribeiro,et al.  Reactive GRASP: An Application to a Matrix Decomposition Problem in TDMA Traffic Assignment , 2000, INFORMS J. Comput..

[23]  E. Burke,et al.  A Randomised Iterative Improvement Algorithm with Composite Neighbourhood Structures for University Course Timetabling , 2005 .

[24]  Philipp Kostuch,et al.  The University Course Timetabling Problem with a Three-Phase Approach , 2004, PATAT.

[25]  Luca Di Gaspero,et al.  Multi-neighbourhood Local Search with Application to Course Timetabling , 2002, PATAT.