Hybridising heuristics within an estimation distribution algorithm for examination timetabling

Rong Qu, Nam Pham, Ruibin Bai, Graham Kendall

Research output: Journal PublicationArticlepeer-review

28 Citations (Scopus)

Abstract

This paper presents a hybrid hyper-heuristic approach based on estimation distribution algorithms. The main motivation is to raise the level of generality for search methodologies. The objective of the hyper-heuristic is to produce solutions of acceptable quality for a number of optimisation problems. In this work, we demonstrate the generality through experimental results for different variants of exam timetabling problems. The hyper-heuristic represents an automated constructive method that searches for heuristic choices from a given set of low-level heuristics based only on non-domain-specific knowledge. The high-level search methodology is based on a simple estimation distribution algorithm. It is capable of guiding the search to select appropriate heuristics in different problem solving situations. The probability distribution of low-level heuristics at different stages of solution construction can be used to measure their effectiveness and possibly help to facilitate more intelligent hyper-heuristic search methods.

Original languageEnglish
Pages (from-to)679-693
Number of pages15
JournalApplied Intelligence
Volume42
Issue number4
DOIs
Publication statusPublished - 1 Jun 2015

Keywords

  • Estimation distribution algorithm
  • Exam timetabling
  • Graph colouring
  • Hyper-heuristic

ASJC Scopus subject areas

  • Artificial Intelligence

Fingerprint

Dive into the research topics of 'Hybridising heuristics within an estimation distribution algorithm for examination timetabling'. Together they form a unique fingerprint.

Cite this