Results 1  10
of
17
Automatic heuristic generation with genetic programming: Evolving a jackofalltrades or a master of one
 GENETIC AND EVOLUTIONARY COMPUTATION CONFERENCE, GECCO 2007, PROCEEDINGS
, 2007
"... It is possible to argue that online bin packing heuristics should be evaluated by using metrics based on their performance over the set of all bin packing problems, such as the worst case or average case performance. However, this method of assessing a heuristic would only be relevant to a user who ..."
Abstract

Cited by 25 (9 self)
 Add to MetaCart
It is possible to argue that online bin packing heuristics should be evaluated by using metrics based on their performance over the set of all bin packing problems, such as the worst case or average case performance. However, this method of assessing a heuristic would only be relevant to a user who employs the heuristic over a set of problems which is actually representative of the set of all possible bin packing problems. On the other hand, a real world user will often only deal with packing problems that are representative of a particular subset. Their piece sizes will all belong to a particular distribution. The contribution of this paper is to show that a Genetic Programming system can automate the process of heuristic generation and produce heuristics that are humancompetitive over a range of sets of problems, or which excel on a particular subset. We also show that the choice of training instances is vital in the area of automatic heuristic generation, due to the tradeoff between the performance and generality of the heuristics generated and their applicability to new problems.
An experimental study on hyperheuristics and exam timetabling
 Proceedings of the 6th International Conference on Practice and Theory of Automated Timetabling
, 2006
"... Abstract. Hyperheuristics are proposed as a higher level of abstraction as compared to the metaheuristics. Hyperheuristic methods deploy a set of simple heuristics and use only nonproblemspecific data, such as, fitness change or heuristic execution time. A typical iteration of a hyperheuristic a ..."
Abstract

Cited by 20 (9 self)
 Add to MetaCart
Abstract. Hyperheuristics are proposed as a higher level of abstraction as compared to the metaheuristics. Hyperheuristic methods deploy a set of simple heuristics and use only nonproblemspecific data, such as, fitness change or heuristic execution time. A typical iteration of a hyperheuristic algorithm consists of two phases: heuristic selection method and move acceptance. In this paper, heuristic selection mechanisms and move acceptance criteria in hyperheuristics are analyzed in depth. Seven heuristic selection methods, and five acceptance criteria are implemented. The performance of each selection and acceptance mechanism pair is evaluated on fourteen wellknown benchmark functions and twentyone exam timetabling problem instances. 1
A Comprehensive Analysis of Hyperheuristics
"... Abstract. Metaheuristics such as simulated annealing, genetic algorithms and tabu search have been successfully applied to many difficult optimization problems for which no satisfactory problem specific solution exists. However, expertise is required to adopt a metaheuristic for solving a problem i ..."
Abstract

Cited by 19 (11 self)
 Add to MetaCart
Abstract. Metaheuristics such as simulated annealing, genetic algorithms and tabu search have been successfully applied to many difficult optimization problems for which no satisfactory problem specific solution exists. However, expertise is required to adopt a metaheuristic for solving a problem in a certain domain. Hyperheuristics introduce a novel approach for search and optimization. A hyperheuristic method operates on top of a set of heuristics. The most appropriate heuristic is determined and applied automatically by the technique at each step to solve a given problem. Hyperheuristics are therefore assumed to be problem independent and can be easily utilized by nonexperts as well. In this study, a comprehensive analysis is carried out on hyperheuristics. The best method is tested against genetic and memetic algorithms on fourteen benchmark functions. Additionally, new hyperheuristic frameworks are evaluated for questioning the notion of problem independence. 1.
A simulated annealing hyperheuristic methodology for flexible decision support
, 2007
"... One of the main motivations for investigating hyperheuristic methodologies is to provide a more general search framework than is currently available. Most of the current search techniques represent approaches that are largely adapted for specific search problems (and, in some cases, even specific ..."
Abstract

Cited by 14 (7 self)
 Add to MetaCart
One of the main motivations for investigating hyperheuristic methodologies is to provide a more general search framework than is currently available. Most of the current search techniques represent approaches that are largely adapted for specific search problems (and, in some cases, even specific problem instances). There are many realworld scenarios where the development of such bespoke systems is entirely appropriate. However, there are other situations where it would be beneficial to have methodologies which are more generally applicable to more problems. One of our motivating goals is to underpin the development of more flexible search methodologies that can be easily and automatically employed on a broader range of problems than is currently possible. Almost all the heuristics that have appeared in the literature have been designed and selected by humans. In this paper, we investigate a simulated annealing hyperheuristic methodology which operates on a search space of heuristics and which employs a stochastic heuristic selection strategy and a shortterm memory. The generality and performance of the proposed algorithm is demonstrated over a large number of benchmark data sets drawn from three very different and difficult (NPhard) problems: nurse rostering, university course timetabling and onedimensional bin packing. Experimental results show that the proposed hyperheuristic is able to achieve significant performance improvements over a recently proposed tabu search hyperheuristic without lowering the level of generality. We
Linear genetic programming of parsimonious metaheuristics
 2007 IEEE CEC
, 2007
"... Abstract — We use a form of grammarbased linear Genetic Programming (GP) as a hyperheuristic, i.e., a search heuristic on the space of heuristics. This technique is guided by domainspecific languages that one designs taking inspiration from elementary components of specialised heuristics and metahe ..."
Abstract

Cited by 13 (6 self)
 Add to MetaCart
Abstract — We use a form of grammarbased linear Genetic Programming (GP) as a hyperheuristic, i.e., a search heuristic on the space of heuristics. This technique is guided by domainspecific languages that one designs taking inspiration from elementary components of specialised heuristics and metaheuristics for a domain. We demonstrate this approach for travelingsalesperson problems for which we test different languages, including one containing a looping construct. Experimentation with benchmark instances from the TSPLIB shows that the GP hyperheuristic routinely and rapidly produces parsimonious metaheuristics that find tours whose lengths are highly competitive with the best realvalued lengths from literature. I.
An Ant Algorithm Hyperheuristic for the Project Presentation Scheduling Problem
 In: Proceedings of the Congress on Evolutionary Computation 2005 (CEC’05). Volume 3
, 2005
"... Ant algorithms have generated significant research interest within the search/optimisation community in recent years. Hyperheuristic research is concerned with the development of "heuristics to choose heuristics" in an attempt to raise the level of generality at which optimisation systems can operat ..."
Abstract

Cited by 10 (4 self)
 Add to MetaCart
Ant algorithms have generated significant research interest within the search/optimisation community in recent years. Hyperheuristic research is concerned with the development of "heuristics to choose heuristics" in an attempt to raise the level of generality at which optimisation systems can operate. In this paper the two are brought together. An investigation of the ant algorithm as a hyperheuristic is presented and discussed. The results are evaluated against other hyperheuristic methods, when applied to a real world scheduling problem.
The Scalability of Evolved On Line Bin Packing Heuristics
"... The on line bin packing problem concerns the packing of pieces into the least number of bins possible, as the pieces arrive in a sequential fashion. In previous work, we used genetic programming to evolve heuristics for this problem, which beat the human designed ‘best fit’ algorithm. Here we exami ..."
Abstract

Cited by 9 (5 self)
 Add to MetaCart
The on line bin packing problem concerns the packing of pieces into the least number of bins possible, as the pieces arrive in a sequential fashion. In previous work, we used genetic programming to evolve heuristics for this problem, which beat the human designed ‘best fit’ algorithm. Here we examine the performance of the evolved heuristics on larger instances of the problem, which contain many more pieces than the problem instances used in training. In previous work, we concluded that we could confidently apply our heuristics to new instances of the same class of problem. Now we can make the additional claim that we can confidently apply our heuristics to problems of much larger size, not only without deterioration of solution quality, but also within a constant factor of the performance obtained by ‘best fit’. Interestingly, our evolved heuristics respond to the number of pieces in a problem instance although they have no explicit access to that information. We also comment on the important point that, when solutions are explicitly constructed for single problem instances, the size of the search space explodes. However, when working in the space of algorithmic heuristics, the distribution of functions represented in the search space reaches some limiting distribution and therefore the combinatorial explosion can be controlled.
A Reinforcement Learning – GreatDeluge Hyperheuristic for Examination
"... Hyperheuristics are identified as the methodologies that search the space generated by a finite set of low level heuristics for solving difficult problems. One of the iterative hyperheuristic frameworks requires a single candidate solution and multiple perturbative low level heuristics. An initial ..."
Abstract

Cited by 6 (6 self)
 Add to MetaCart
Hyperheuristics are identified as the methodologies that search the space generated by a finite set of low level heuristics for solving difficult problems. One of the iterative hyperheuristic frameworks requires a single candidate solution and multiple perturbative low level heuristics. An initially generated complete solution goes through two successive processes; heuristic selection and move acceptance until a set of termination criteria is satisfied. A goal of the hyperheuristic research is to create automated techniques that are applicable to wide range of problems with different characteristics. Some previous studies show that different combinations of heuristic selection and move acceptance as hyperheuristic components might yield different performances. This study investigates whether learning heuristic selection can improve the performance of a great deluge based hyperheuristic using an examination timetabling problem as a case study.
Hybrid Variable Neighborhood HyperHeuristics For Exam Timetabling
"... Introduction Exam timetabling is one of the most important administrative activities in universities (e.g. [4], [8], and [19]). Such problems have been the subject of significant research e#ort across Artificial Intelligence research and Operational Research since the 1960s. A general exam timetab ..."
Abstract

Cited by 3 (2 self)
 Add to MetaCart
Introduction Exam timetabling is one of the most important administrative activities in universities (e.g. [4], [8], and [19]). Such problems have been the subject of significant research e#ort across Artificial Intelligence research and Operational Research since the 1960s. A general exam timetabling problem consists of assigning a set of exams into a limited number of timeslots while satisfying a set of constraints: Hard constraints that cannot be violated in any circumstances; and Soft constraints which should be satisfied as much as is possible. The area of metaheuristics [12], [17] has demonstrated success in developing stateoftheart timetabling approaches. However these approaches are usually finetuned on particular problems and thus cannot be easily applied to other problems. Hyperheuristics [3] have been employed recently with some success and are motivated by the aim of improving the generality of search methodologies to facilitate the automatic generation of solution t
Toward Subheuristic Search
 In 2008 IEEE WCCI
, 2008
"... Abstract — In previous work, we have introduced an effective, resourceefficient and selfadapting hyperheuristic that uses Genetic Programming (GP) as its method of search in the space of domainspecific metaheuristics. GP employs userprovided, local heuristics from which it produces these metaheu ..."
Abstract

Cited by 1 (1 self)
 Add to MetaCart
Abstract — In previous work, we have introduced an effective, resourceefficient and selfadapting hyperheuristic that uses Genetic Programming (GP) as its method of search in the space of domainspecific metaheuristics. GP employs userprovided, local heuristics from which it produces these metaheuristics (MHs). Here, we show that the hyperheuristic performs even better when working at the subheuristic level, i.e., when building MHs from generic components and specific elementary operations. In particular, this approach supports efficiency of the better MHs. Specifically, these MHs do not excessively iterate local search steps, i.e., their good performance comes from smart patterns of calls of the provided, basic components. Also, a moderate reduction of the maximum allowed MH size does not reduce performance significantly. I.