Hyper-heuristics

Ross, Peter

doi:10.1007/978-1-4614-6940-7_20

Cited by 32 publications

(37 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Both selection and generation hyper-heuristics can be further categorized into construction or perturbation heuristics. More on hyper-heuristics can be found in [3,13,16]. Selection hyperheuristics have been well studied, and gave rise to the CHeSC 2011 competition 3 ; further details of this can be found in [7,12] and at the CHeSC website, and of the winning hyper-heuristic by Misir et al in [11].…”

Section: Introduction and Related Workmentioning

confidence: 99%

Generalizing Hyper-heuristics via Apprenticeship Learning

Asta

Özcan

Parkes

et al. 2013

Evolutionary Computation in Combinatorial Optimization

View full text Add to dashboard Cite

Abstract. An apprenticeship-learning-based technique is used as a hyperheuristic to generate heuristics for an online combinatorial problem. It observes and learns from the actions of a known-expert heuristic on small instances, but has the advantage of producing a general heuristic that works well on other larger instances. Specifically, we generate heuristic policies for online bin packing problem by using expert near-optimal policies produced by a hyper-heuristic on small instances, where learning is fast. The "expert" is a policy matrix that defines an index policy, and the apprenticeship learning is based on observation of the action of the expert policy together with a range of features of the bin being considered, and then applying a k-means classification. We show that the generated policy often performs better than the standard best-fit heuristic even when applied to instances much larger than the training set.

show abstract

Section: Introduction and Related Workmentioning

confidence: 99%

Generalizing Hyper-heuristics via Apprenticeship Learning

Asta

Özcan

Parkes

et al. 2013

Evolutionary Computation in Combinatorial Optimization

View full text Add to dashboard Cite

show abstract

“…Metaheuristic techniques are often very effective; however, there can be some reluctance to use them for money-critical problems. Practitioners in industry often favor the use of very simple and readily understandable methods even if they deliver relatively inferior results (Ross 2005). Among the main criticisms of stochastic-based problem solving techniques are: (1) the fact that they involve some randomness and unpredictability, so that identical runs may deliver very different results; (2) there is little understanding about their average-and worst-case behavior (Ross and Marín-Blázquez 2005); (3) solutions quality greatly depend on a good parameter choice; and (4) the parameter tunning task requires time, knowledge and experience about the problem domain and properties which makes metaheuristics problemspecific solution methods that can be developed and deployed only by experts (Ross 2005;Bilgin et al 2006).…”

Section: The 2d Irregular Bin Packing Problemmentioning

confidence: 99%

“…In bin packing, especially in the 1D case, instances with many small items are not hard, since the small items can be employed as sand to fill up the remaining space when large items were packed. Difficulty arises when most of the pieces have an area that is a significant fraction of the total object area, for example at least 20 % of the object area, so that the challenge is to find the subset among a large number of pieces to be placed in a given object (Ross 2005). We hypothesize that the DJD heuristic is intended for these kind of hard instances; because, exactly as stated, it works well in many problems known to be hard, but it fails in other types of problem.…”

Section: The Djd Heuristicmentioning

confidence: 99%

“…However, DJD will first fill a bin until it contains 334 items (just over one-third) and then add just three more items into the bin, so the bin will contain 337 items. Thus, 30 bins will be needed (337 × 29 = 9773) (Ross 2005), a solution far from optimality. The obvious remedy to this situation is to keep trying to place item combinations until no single item can be placed.…”

Section: The Djd Heuristicmentioning

confidence: 99%

See 1 more Smart Citation

An effective heuristic for the two-dimensional irregular bin packing problem

et al. 2013

View full text Add to dashboard Cite

This paper proposes an adaptation, to the two-dimensional irregular bin packing problem of the Djang and Finch heuristic (DJD), originally designed for the one-dimensional bin packing problem. In the two-dimensional case, not only is it the case that the piece's size is important but its shape also has a significant influence. Therefore, DJD as a selection heuristic has to be paired with a placement heuristic to completely construct a solution to the underlying packing problem. A successful adaptation of the DJD requires a routine to reduce computational costs, which is also proposed and successfully tested in this paper. Results, on a wide variety of instance types with convex polygons, are found to be significantly better than those produced by more conventional selection heuristics.

show abstract

“…In this case, a hyper-heuristic was defined as a high-level approach that, given a particular problem instance and a number of atomic heuristics, selects and applies an appropriate heuristic at each decision point [14,96]. This definition of hyper-heuristics was also expanded later to refer to an automated methodology for selecting or generating heuristics to solve hard computational search problems [16].…”

Section: Introductionmentioning

confidence: 99%

Contrasting meta-learning and hyper-heuristic research: the role of evolutionary algorithms

Pappa

Ochoa

Hyde

et al. 2013

Genet Program Evolvable Mach

View full text Add to dashboard Cite

The fields of machine meta-learning and hyper-heuristic optimisation have developed mostly independently of each other, although evolutionary algorithms (particularly genetic programming) have recently played an important role in the development of both fields. Recent work in both fields shares a common goal, that of automating as much of the algorithm design process as possible. In this paper we first provide a historical perspective on automated algorithm design, and then we discuss similarities and differences between meta-learning in the field of supervised machine learning (classification) and hyper-heuristics in the field of optimisation. This discussion focuses on the dimensions of the problem space, the algorithm space and the performance measure, as well as clarifying important issues related to different levels of automation and generality in both fields. We also discuss important research directions, challenges and foundational issues in meta-learning and hyper-heuristic research. It is important to emphasize that this paper is not a survey, as several surveys on the areas of meta-learning and hyper-heuristics (separately) have been previously published. The main contribution of the paper is to contrast meta-learning and hyper-heuristics methods and concepts, in order to 123Genet Program Evolvable Mach DOI 10.1007/s10710-013-9186-9 promote awareness and cross-fertilisation of ideas across the (by and large, nonoverlapping) different communities of meta-learning and hyper-heuristic researchers. We hope that this cross-fertilisation of ideas can inspire interesting new research in both fields and in the new emerging research area which consists of integrating those fields.

show abstract

Hyper-heuristics

Cited by 32 publications

References 32 publications

Generalizing Hyper-heuristics via Apprenticeship Learning

Generalizing Hyper-heuristics via Apprenticeship Learning

An effective heuristic for the two-dimensional irregular bin packing problem

Contrasting meta-learning and hyper-heuristic research: the role of evolutionary algorithms

Contact Info

Product

Resources

About