Random Subsets Support Learning a Mixture of Heuristics

Petrovic, Smiljana; Epstein, Susan L.

doi:10.1142/s0218213008004023

Cited by 13 publications

(7 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are several examples in the literature of empirical research concerning subsets of low level heuristics such as Petrovic and Epstein (2008), and Soria-Alcaraz et al (2017), and from a purely theoretical perspective (Lehre and Özcan 2013). For example, in Petrovic and Epstein (2008) subsets of heuristics are randomly selected from a large pool of available heuristics. Each subset is evaluated on a number of benchmark problems, and a learning algorithm is used to determine weights for the elements of the subset.…”

Section: Heuristic Subsequencesmentioning

confidence: 99%

An analysis of heuristic subsequences for offline hyper-heuristic learning

Yates

Keedwell

2019

J Heuristics

View full text Add to dashboard Cite

A selection hyper-heuristic is used to minimise the objective functions of a wellknown set of benchmark problems. The resulting sequences of low level heuristic selections and objective function values are used to generate a database of heuristic selections. The sequences in the database are broken down into subsequences and the mathematical concept of a logarithmic return is used to discriminate between "effective" subsequences, which tend to decrease the objective value, and "disruptive" subsequences, which tend to increase the objective value. These subsequences are then employed in a sequenced based hyper-heuristic and evaluated on an unseen set of benchmark problems. Empirical results demonstrate that the "effective" subsequences perform significantly better than the "disruptive" subsequences across a number of problem domains with 99% confidence. The identification of subsequences of heuristic selections that can be shown to be effective across a number of problems or problem domains could have important implications for the design of future sequence based hyper-heuristics.

show abstract

Section: Heuristic Subsequencesmentioning

confidence: 99%

An analysis of heuristic subsequences for offline hyper-heuristic learning

Yates

Keedwell

2019

J Heuristics

View full text Add to dashboard Cite

show abstract

“…The most extreme case in this representation would be an algorithm containing an infinite switch-case statement containing all the problems and the best heuristic for each one of them. Hyper-heuristics are closely related to dynamic algorithm portfolios [25,30] in the way they work. Both algorithm portfolios and hyper-heuristics select from existing methods to apply the most suitable one at different stages of the search.…”

Section: Hyper-heuristicsmentioning

confidence: 99%

“…More recent studies on the combination of heuristics for CSPs include the work done by Petrovic and Epstein [30], who studied the idea of combining various heuristics to produce mixtures that work well on some sets of CSP instances. Their approach bases their decisions on random sets of performance-based weighted criteria that variable and value ordering heuristics use to make their decisions.…”

Section: Hyper-heuristicsmentioning

confidence: 99%

A Neuro-evolutionary Hyper-heuristic Approach for Constraint Satisfaction Problems

2015

View full text Add to dashboard Cite

Constraint satisfaction problems represent an important topic of research due to their multiple applications in various areas of study. The most common way to solve this problem involves the use of heuristics that guide the search into promising areas of the space. In this article, we present a novel way to combine the strengths of distinct heuristics to produce solution methods that perform better than such heuristics on a wider range of instances. The methodology proposed produces neural networks that represent hyper-heuristics for variable ordering in constraint satisfaction problems. These neural networks are generated and trained by running a genetic algorithm that has the task of evolving the topology of the networks and some of their learning parameters. The results obtained suggest that the produced neural networks represent a feasible alternative for coding hyper-heuristics that control the use of different heuristics in such a way that the cost of the search is minimized.

show abstract

“…Also, algorithm portfolios for Constraint programming have been successfully studied before [13]. Petrovic and Eipstein [25] studied the idea of combining various heuristics to produce mixtures that work well on particular classes of instances. More recent studies about the dynamic combination of heuristics applied to CSP include the work done by Terashima-Marín et al [35], who proposed an evolutionary framework to generate hyper-heuristics for variable ordering in CSP and the research developed by Bittle and Fox [5] who presented a hyper-heuristic approach for variable and value ordering for CSP based on a symbolic cognitive architecture augmented with case based reasoning as the machine learning mechanism for their hyper-heuristics.…”

Section: Hyper-heuristicsmentioning

confidence: 99%

Improving the performance of vector hyper-heuristics through local search

Ortíz-Bayliss

Terashima-Marín

Conant-Pablos

et al. 2012

Proceedings of the 14th Annual Conference on Genetic and Evolutionary Computation

View full text Add to dashboard Cite

Hyper-heuristics are methodologies that allow us to selectively apply the most suitable heuristic given the properties of the problem at hand. They can be applied in CSP in different ways, but one way which has received attention in recent years is variable ordering by using hyper-heuristics. To select the next variable, a set of heuristics exist and the hyper-heuristic decides, considering the features that describe the instance at hand, which heuristic is more suitable to be applied at the moment. This paper explores a hyperheuristic model for variable ordering within CSP based on vector hyper-heuristics. Each hyper-heuristic is represented as a set of vectors that maps instance features to heuristics. These vector hyper-heuristics are constructed by going into a local search method that modifies the hyper-heuristics. The results suggest that the approach is able to combine the strengths of different heuristics to perform well on a wide range of instances and compensate for their weaknesses on specific instances, resulting in an improvement in the performance of the search compared against the heuristics applied in isolation.

show abstract

Random Subsets Support Learning a Mixture of Heuristics

Cited by 13 publications

References 16 publications

An analysis of heuristic subsequences for offline hyper-heuristic learning

An analysis of heuristic subsequences for offline hyper-heuristic learning

A Neuro-evolutionary Hyper-heuristic Approach for Constraint Satisfaction Problems

Improving the performance of vector hyper-heuristics through local search

Contact Info

Product

Resources

About