Graham Kendall
Various Images

Professor Graham Kendall

University of Nottingham, UK

I am a Professor of Computer Science at the University of Nottingham (UK). I am currently the Vice-Provost (Research and Knowledge Transfer) at our campus in Malaysia. I am a member of the Automated Scheduling, Optimisation and Planning (ASAP) Research Group. My interests include Operational Research, Evolutionary Computing, Scheduling (particularly sports scheduling), Cutting and Packing, Timetabling and Games (both games in the usual sense of the word as well as mathematical games such as the Iterated Prisoners Dilemma).

Latest Blog Post

How to get ants to solve a chess problem

Random Blog Post

Some of the Challenges in Parsing Bibtex Authors

News

I am a director of a spin out company that specialises in Nesting Software
http://bit.ly/i9wYHy

Publication

Constructing Initial Neighbourhoods to Identify Critical Constraints
http://bit.ly/h3xfnd

Publication

Construction of Initial Neighborhoods for a Course Scheduling Problem Using Tiling
http://bit.ly/dMXRME

Publication

A Strategy with Novel Evolutionary Features for the Iterated Prisoner's Dilemma
http://bit.ly/eURggX

Publication

The implementation of a novel, bio-inspired, robotic security system
http://bit.ly/1dvguUh

Graham Kendall: Details of Requested Publication


Citation

Cowling, P; Kendall, G and Soubeiga, E A Parameter-Free Hyperheuristic for Scheduling a Sales Summit. In Proceedings of the 4th Metahuristics International Conference (MIC 2001), pages 127-131, 16-20 July 2001, Porto Portugal, 2001.

There is not an abstract for this paper, so we have used the introduction as the abstract.


Abstract

Personnel scheduling involves scheduling people to timeslots and possibly locations. This problem remains the subject of much research interest with a survey in every decade since the 1970ís. Several models including Constraint Satisfaction Problem (CSP), mathematical programming (linear programming, 0-1 integer programming etc.), and different solution methods (exact or heuristic based) have been proposed to tackle this problem. Also several applications have been reported in the literature including applications for hospital personnel, educational institution personnel and others. However the heuristic methods developed for a particular personnel scheduling problem may not perform well if applied to a different problem. Heuristic and metaheuristic approaches tend to be knowledge rich, requiring substantial expertise in both the problem domain and appropriate heuristic techniques. It is in this context that we proposed a hyperheuristic approach as a heuristic that operates at a higher level of abstraction than current metaheuristic approaches. The hyperheuristic has no problem-specific knowledge. It manages a set of simple, knowledge-poor, low-level heuristics. At any given time the hyperheuristic must choose which low-level heuristic to call. The hyperheuristic interacts with the low-level heuristics but is only allowed to communicate non problem-specific information such as CPU time and the change in the evaluation function. Problem-specific domain knowledge is prohibited from passing through the hyperheuristic/low-level heuristic interface. To allow the hyperheuristic to operate, a choice function is defined which adaptively ranks the low-level heuristics. The choice function provides guidance to the hyperheuristic by indicating which low-level heuristic should be applied next based upon the historical performance of the heuristics and the region of the search space currently under exploration. In a choice function was defined as a weighted sum of three factors expressing the recent improvement (change in the evaluation function) produced by each of the heuristics, the recent improvement produced by a consecutive pair of heuristics and the time since the heuristic had been called. Although the hyperheuristic produced good results, these results depended on the values of the parameters, and such values were empirically determined by simply choosing the best settings after many trials. Empirically setting the parameters increases the domain-specific knowledge of the hyperheuristic, which is not desirable. This paper is concerned with the development of a mechanism for automatically setting the parameters as well as refining the choice function. To do so we apply the hyperheuristic to a real-world personnel scheduling problem, that of scheduling a sales summit.


pdf

You can download the pdf of this publication from here


doi

This publication does not have a doi, so we cannot provide a link to the original source

What is a doi?: A doi (Document Object Identifier) is a unique identifier for sicientific papers (and occasionally other material). This provides direct access to the location where the original article is published using the URL http://dx.doi/org/xxxx (replacing xxx with the doi). See http://dx.doi.org/ for more information



URL

This pubication does not have a URL associated with it.

The URL is only provided if there is additional information that might be useful. For example, where the entry is a book chapter, the URL might link to the book itself.


Bibtex

@INPROCEEDINGS{cks2001a, author = {P. Cowling and G. Kendall and E. Soubeiga},
title = {A Parameter-Free Hyperheuristic for Scheduling a Sales Summit},
booktitle = {Proceedings of the 4th Metahuristics International Conference (MIC 2001)},
year = {2001},
pages = {127--131},
address = {16-20 July 2001, Porto Portugal},
note = {There is not an abstract for this paper, so we have used the introduction as the abstract.},
abstract = {Personnel scheduling involves scheduling people to timeslots and possibly locations. This problem remains the subject of much research interest with a survey in every decade since the 1970ís. Several models including Constraint Satisfaction Problem (CSP), mathematical programming (linear programming, 0-1 integer programming etc.), and different solution methods (exact or heuristic based) have been proposed to tackle this problem. Also several applications have been reported in the literature including applications for hospital personnel, educational institution personnel and others. However the heuristic methods developed for a particular personnel scheduling problem may not perform well if applied to a different problem. Heuristic and metaheuristic approaches tend to be knowledge rich, requiring substantial expertise in both the problem domain and appropriate heuristic techniques. It is in this context that we proposed a hyperheuristic approach as a heuristic that operates at a higher level of abstraction than current metaheuristic approaches. The hyperheuristic has no problem-specific knowledge. It manages a set of simple, knowledge-poor, low-level heuristics. At any given time the hyperheuristic must choose which low-level heuristic to call. The hyperheuristic interacts with the low-level heuristics but is only allowed to communicate non problem-specific information such as CPU time and the change in the evaluation function. Problem-specific domain knowledge is prohibited from passing through the hyperheuristic/low-level heuristic interface. To allow the hyperheuristic to operate, a choice function is defined which adaptively ranks the low-level heuristics. The choice function provides guidance to the hyperheuristic by indicating which low-level heuristic should be applied next based upon the historical performance of the heuristics and the region of the search space currently under exploration. In a choice function was defined as a weighted sum of three factors expressing the recent improvement (change in the evaluation function) produced by each of the heuristics, the recent improvement produced by a consecutive pair of heuristics and the time since the heuristic had been called. Although the hyperheuristic produced good results, these results depended on the values of the parameters, and such values were empirically determined by simply choosing the best settings after many trials. Empirically setting the parameters increases the domain-specific knowledge of the hyperheuristic, which is not desirable. This paper is concerned with the development of a mechanism for automatically setting the parameters as well as refining the choice function. To do so we apply the hyperheuristic to a real-world personnel scheduling problem, that of scheduling a sales summit.},
keywords = {hyper-heuristic, hyperheuristic, scheduling, personnel scheduling, rostering},
timestamp = {2007.03.29},
webpdf = {http://www.graham-kendall.com/papers/cks2001a.pdf} }