A Grouping Genetic Algorithm Using Linear Linkage Encoding for Bin Packing

Proceedings of the 13th Annual Conference on Genetic and Evolutionary Computation

Parkes

2011

Self Cite

Online bin-packing is a well-known problem in which immediate decisions must be made about the placement of items with various sizes into fixed capacity bins. The associated decisions can be based on an index policy in which each decision option is independently given a value and the highest value choice is selected. In this paper, we represent such heuristics for online bin packing as a simple matrix of scores. We then use a genetic algorithm to search for matrices giving good performance. This might be regarded as parameter tuning of the packing heuristic but in which a fine-grained representation is used and so the number of parameters is much larger than in standard parameter tuning. The evolved matrices perform better than the standard heuristics. They also reveal interesting structures and so have impact on questions of how heuristic score functions should be represented and what structure they might be expected to exhibit.

Section: Bin-packingmentioning

confidence: 99%

Policy matrix evolution for generation of heuristics

Proceedings of the 13th Annual Conference on Genetic and Evolutionary Computation

Parkes

2011

Self Cite

“…GA performs significantly better than the tuning approach on the latter two instances. The policy generated via tuning performs very much similar to the one generated by GA for U BP (6,2,3). When the AL approach is compared to the policy matrices tuned via the racing algorithm, we have observed that both proposed approaches outperform AL over four out of six instance generators.…”

Section: B Experimental Resultsmentioning

confidence: 87%

“…Both selection and generation hyper-heuristics can be further categorized into construction or perturbation heuristics. More on hyper-heuristics can be found in [5], [6] and [7]. Selection hyper-heuristics have been well studied, giving rise to a challenge ( [8], [9]) and the winning online learning hyper-heuristic was provided by Misir et al [10].…”

Section: Introductionmentioning

confidence: 99%

Heuristic generation via parameter tuning for online bin packing

Yarimcam

Asta

2014 IEEE Symposium on Evolving and Autonomous Learning Systems (EALS)

et al. 2014

Abstract-Online bin packing requires immediate decisions to be made for placing an incoming item one at a time into bins of fixed capacity without causing any overflow. The goal is to maximise the average bin fullness after placement of a long stream of items. A recent work describes an approach for solving this problem based on a 'policy matrix' representation in which each decision option is independently given a value and the highest value option is selected. A policy matrix can also be viewed as a heuristic with many parameters and then the search for a good policy matrix can be treated as a parameter tuning process. In this study, we show that the Irace parameter tuning algorithm produces heuristics which outperform the standard human designed heuristics for various instances of the online bin packing problem.

“…Hyper-heuristics are high level search and optimisation methods which explore the space formed by low level heuristics or heuristic components for solving complex problems (Burke et al, 2003;Ross, 2005;Özcan et al, 2008;Chakhlevitch and Cowling, 2008;Burke et al, 2013). Burke et al (2010) classified hyper-heuristics into two main categories; methodologies to select or generate heuristics.…”

Section: Hyper-heuristics For Bin Packingmentioning

confidence: 99%

CHAMP: Creating heuristics via many parameters for online bin packing

Asta

Expert Systems with Applications

Parkes

2016

Self Cite

The online bin packing problem is a well-known bin packing variant which requires immediate decisions to be made for the placement of a lengthy sequence of arriving items of various sizes one at a time into fixed capacity bins without any overflow. The overall goal is maximising the average bin fullness. We investigate a 'policy matrix' representation which assigns a score for each decision option independently and the option with the highest value is chosen for one dimensional online bin packing. A policy matrix might also be considered as a heuristic with many parameters, where each parameter value is a score.We hence investigate a framework which can be used for creating heuristics via many parameters. The proposed framework combines a Genetic Algorithm optimiser, which searches the space of heuristics in policy matrix form, and an online bin packing simulator, which acts as the evaluation function. The empirical results indicate the success of the proposed approach, providing the best solutions for almost all item sequence generators used during the experiments.We also present a novel fitness landscape analysis on the search space of policies. This study hence gives evidence of the potential for automated discovery by intelligent systems of powerful heuristics for online problems; reducing the need for expensive use of human expertise.