Multivariate convergence-targeted operator for the genetic algorithm

Authors

DOI:

https://doi.org/10.20535/SRIT.2308-8893.2017.4.10

Keywords:

machine learning, genetic algorithm, Pareto Front, principle component analysis, transport particle simulations

Abstract

Optimization of complex particle transport simulation packages could be managed using genetic algorithms as a tuning instrument for learning statistics and behavior of multi-objective optimisation functions. Combination of genetic algorithm and unsupervised machine learning could significantly increase convergence of algorithm to true Pareto Front (PF). We tried to apply specific multivariate analysis operator that can be used in case of expensive fitness function evaluations, in order to speed-up the convergence of the "black-box" optimization problem. The results delivered in the article shows that current approach could be used for any type of genetic algorithm and deployed as a separate genetic operator.

Author Biographies

Oksana Shadura, The Bogolyubov Institute of Theoretical Physics NAS of Ukraine, Kyiv

Oksana Shadura,

PhD candidate at Institute for Applied System Analysis NTUU "KPI", engineer of the Bogolyubov Institute of Theoretical Physics NAS of Ukraine, Kyiv, Ukraine, and Doctoral Student at CERN, Switzerland. Graduated from the Kiev Polytechnic Institute in 2012. Field of research: performance optimisation, machine learning, HPC, transport particle simulations.

Anatoly I. Petrenko, The Institute for Applied System Analysis at the Igor Sikorsky Kyiv Polytechnic Institute, Kyiv

Petrenko Anatoly Ivanovich,

Professor, Doctor of Technical Sciences (DSc), Head of the Department of System Design and the Department of information Resources, the Institute for Applied System Analysis at the Igor Sikorsky Kyiv Polytechnic Institute, and National Academy of Sciences (NAS) of Ukraine. He graduated from the Kiev Polytechnic Institute in 1957. Field of research: service-oriented computing to solve practical interdisciplinary problems in the intellectual environment; organization of distributed program-technical complexes for networked collective design based on grid / cloud computing.

Sergiy Ya. Svistunov, The Bogolyubov Institute of Theoretical Physics NAS of Ukraine, Kyiv

Sergiy Yakovych Svistunov,

PhD, Head of the Department for Computer Maintenance of the Bogolyubov Institute for Theoretical Physics of NAS of Ukraine, assistant professor Educational-scientific complex "The Institute for Applied Systems Analysis" NTU "KPI", Kyiv, Ukraine. Graduated from the Kiev Polytechnic Institute in 1980.

Field of research: distributed computing systems, grid/cloud computing.

References

Amadio G. GeantV: from CPU to accelerators / G. Amadio, A. Ananya, J. Apostolakis, A. Arora, M. Bandieramonte, A. Bhattacharyya, C. Bianchini, R. Brun, P. Canal, F. Carminati, L. Duhem, D. Elvira, A. Gheata, M. Gheata, I. Goulas, R. Iope, S. Jun, G. Lima, A. Mohanty, T. Nikitina, M. Novak, W. Pokorski, A. Ribon, R. Sehgal, O. Shadura, S. Vallecorsa, S. Wenzel, Y. Zhang // Journal of Physics: Conference Series. — 2016. — Vol. 762, N 1. — P. 012019.

Allison J. Geant4 developments and applications / J. Allison, K. Amako, J. Apostolakis, H. Araujo, P. Arce Dubois, M. Asai, G. Barrand, R. Capra and others // IEEE Transactions on Nuclear Science. — 2006. — Vol. 53, N 1. — P. 270–278.

Amadio G. The GeantV project: preparing the future of simulation / G. Amadio, J. Apostolakis, M.Bandieramonte, A. Bhattacharyya, C. Bianchini, R. Brun and others // Journal of Physics: Conference Series. — 2015. — Vol. 664, N 7. — P. 072006.

Apostolakis J. Towards a high performance geometry library for particle-detector simulations / J.Apostolakis, M.Bandieramonte, G. Bitzes, R. Brun, P. Canal, F. Carminati and others // Journal of Physics: Conference Series. — 2015. — Vol. 608, N 1. — P. 012023.

Apostolakis J. Adaptive track scheduling to optimize concurrency and vectorization in GeantV / J.Apostolakis, M.Bandieramonte, G. Bitzes, R. Brun, P. Canal, F Carminati and others // Journal of Physics: Conference Series. — 2015. — Vol. 608, N 1. — P. 012003.

Deb K. Scalable Test Problems for Evolutionary Multi-Objective Optimization, Evolutionary Multiobjective Optimization: Theoretical Advances and Applications / K. Deb, L. Thiele, M. Laumanns, E. Zitzler // Advanced Information and Knowledge Processing - Evolutionary Multiobjective Optimization - Springer London. — 2015. — P. 105–145.

Vose M. The Simple Genetic Algorithm: Foundations and Theory / M. Vose // MIT Press. Cambridge. — 1999. — 251 p.

Rowe J.E. Genetic algorithm theory / J.E. Rowe // Proceedings of the 14th Annual Conference Companion on Genetic and Evolutionary Computation (GECCO2012). — New York, USA. — 2012. — P. 917–940.

Rudolph G. Convergence properties of evolutionary algorithms / G. Rudolph // Kovac, Hamburg. — 1997. — 286 p.

Schmitt F. On the Importance of the Second Largest Eigenvalue on the Convergence Rate of Genetic Algorithms / F. Schmitt, F. Rothlauf // Proceedings of the 14th Symposium on Reliable Distributed Systems. — 2001. — P. 559–564.

Cadima J. On Relationships between Uncentered and Column-centered Principal Component Analysis / J. Cadima, I. Jolliffe // Pakistan Journal of Statistics. — 2009. — Vol. 25(4). — P. 473–503.

Honeine P. An eigenanalysis of data centering in machine learning / P. Honeine // Preprint ArXiV ID: 1407.2904. — 2014. — 14 p.

Baker K. Singular value decomposition tutorial / K. Baker. — Available at: https://datajobs.com/data-science-repo/SVD-Tutorial-%5BKirk-Baker%5D.pdf

Deb K. A fast and elitist multiobjective genetic algorithm: NSGA-II / K. Deb, A. Pratap, S. Agarwal, T. Meyarivan // IEEE Transactions on Evolutionary Computation. — 2002. — Vol. 6. — P.182–197.

Haitham S. U-NSGA-III: A unified evolutionary algorithm for single, multiple, and many-objective optimization / S. Haitham, K. Deb // Lecture Notes in Computer Science. — 2015. — Vol. 6019. — P. 34–49.

Annadurai S. Fundamentals of digital image processing / S. Annadurai // Pearson Education India. — 2007. — 440 p.

Hansen N. COCO: A platform for Comparing Continuous Optimizers in a Black-Box Setting / N. Hansen, A. Auger, O. Mersmann, T. Tuar, D. Brockhoff // Preprint ArXiV ID: 1603.08785. — 2016. — 10 p.

Downloads

Published

2017-03-21

Issue

Section

Methods of system analysis and control in conditions of risk and uncertainty