A genetic Gaussian process regression model based on memetic algorithm

Abstract

Gaussian process (GP) has fewer parameters, simple model and output of probabilistic sense, when compared with the methods such as support vector machines. Selection of the hyper-parameters is critical to the performance of Gaussian process model. However, the common-used algorithm has the disadvantages of difficult determination of iteration steps, over-dependence of optimization effect on initial values, and easily falling into local optimum. To solve this problem, a method combining the Gaussian process with memetic algorithm was proposed. Based on this method, memetic algorithm was used to search the optimal hyper parameters of Gaussian process regression (GPR) model in the training process and form MA-GPR algorithms, and then the model was used to predict and test the results. When used in the marine long-range precision strike system (LPSS) battle effectiveness evaluation, the proposed MA-GPR model significantly improved the prediction accuracy, compared with the conjugate gradient method and the genetic algorithm optimization process.

Keywords

Gaussian process / hyper-parameters optimization / memetic algorithm / regression model

References

[1] SunB, YaoH-t, LiuTing. Prediction of short-term wind speed based on Gaussian process [J]. Proceedings of the CSEE, 2012, 32(29): 104-109

[2] SunZ-h, ShenQ-qianStudy of the online modeling problem based on Gaussian process [D], 2011GuangzhouSouth China University of Technology

[3] SuG-s, QuanJiang. A cooperative optimization algorithm based on gaussi an process and particle swarm optimization for optimizing expensive problems [C]. International Joint Conference on Computational Sciences and Optimization, 2009Los AlamitosIEEE Computer Society929-933

[4] SuG-s, ZhangK-s, LuH-bo. Particle swarm optimization in displacement back analysis-coordinated optimization technique of Gaussian process [J]. Rock & Soil Mechanics, 2011, 32(2): 510-515

[5] XuC, LiuB-g, LiuK-y, GuoJ-qi. Intelligent model for analysis & prediction of landslide displacement time series based on particle swarm-Gaussian process regression coupling algorithm [J]. Rock & Soil Mechanics, 2011, 32(6): 1669-1675

[6] LiuK-y, FangY, LiuB-g, XuChong. Genetic Gaussian process regression model for prediction of tunnel surrounding rock deformation [J]. Journal of the China Railway Society, 2011, 33(12): 101-106

[7] XuC, LiuB-g, LiuK-y, GuoJ-qi. Intelligent design of Gaussian process slope angle based on compound kernel function [J]. Rock & Soil Mechanics, 2010, 31(3): 821-826

[8] TenneY, StevenW A F. A framework for memetic optimization using variable global and local surrogate models [J]. Soft Computing, 2009, 13(8/9): 781-793

[9] PadmavathiS, MohitgolchhaS, SeenimohamedA. Memetic algorithm based task scheduling using probabilistic local search [C]. Swarm, Evolutionary, and Memetic Computing, 2012Berlin HeidelbergSpringer224-231

[10] MoscatoP, CarlosCA gentle introduction to memetic algorithms [M], 2003New YorkUS, Springer105-144

[11] BergstraJ, YoshuaB. Random search for hyper-parameter optimization [J]. The Journal of Machine Learning Research, 2012, 13: 281-305

[12] RasmussenC E, ChristopherK IGaussian processes for machine learning [M], 2005Boston: AmericaThe MIT Press

[13] YaoF-tianStudy of hyper-spectral image classification based on gaussian process [D], 2011HangzhouZhejiang University

[14] MoscatoP. On evolution, search, optimization, genetic algorithms and martial arts: Towards memetic algorithms [R]. Caltech Concurrent Computation Program, C 3P Report 826, 1989

[15] MelanieMAn introduction to genetic algorithms [M], 1999Cambridge, EnglandThe MIT Press

[16] LiY, LiuG, LaoS-yang. A genetic algorithm for community detection in complex networks [J]. Journal of Central South University, 2013, 20(5): 1269-1276

[17] ZhaoW-z, WangC-y, YuL-y, ChenTao. Performance optimization of electric power steering based on multi-objective genetic algorithm [J]. Journal of Central South University, 2013, 20(1): 98-104

[18] RodriguezT E, JoseT J. Memetic algorithms for constructing binary covering arrays of strength three [C]. 9th International Conference on Artificial Evolution, 2010Berlin HeidelbergSpringer86-97

[19] HauptR L, WernerD HGenetic algorithms in electro magnetics [M], 2007New JerseyJohn Willy & Sons

[20] GuimaraesF G, CampeloF, IgarashiH, LowtherD A, RamirezJ A. Optimization of cost functions using evolutionary algorithms with local learning and local search [J]. Magnetics, IEEE Transactions on, 2007, 43(4): 1641-1644

[21] MichalewicaZGenetic algorithms data structures evolution programs [M], 1996BerlinSpringer-Verlag