On evolutionary optimization with approximate fitness functions
作者： Markus Olhofer , Bernhard Sendhoff , Yaochu Jin
DOI:
关键词: Fitness approximation 、 Rosenbrock function 、 Evolutionary algorithm 、 Interactive evolutionary computation 、 Mathematics 、 Evolution strategy 、 CMA-ES 、 Evolutionary programming 、 Evolutionary computation 、 Mathematical optimization
摘要: The evaluation of the quality solutions is usually very time-consuming in design optimization. Therefore, time-efficient approximate models can be particularly beneficial for when evolutionary algorithms are applied. In this paper, convergence property an evolution strategy (ES) with neural network based fitness evaluations investigated. It found that algorithm will converge incorrectly if model has false optima. To address problem, two strategies to control process introduced. addition, methods eliminate minima training proposed. effectiveness shown simulation studies on Ackley function and Rosenbrock function.
acm.org 参考文章(21)
Ulrich Hammel, Thomas Bäck, Evolution Strategies on Noisy Functions: How to Improve Convergence Properties parallel problem solving from nature. pp. 159- 168 ,(1994) , 10.1007/3-540-58484-6_260
Markus Olhofer, Bernhard Sendhoff, Toshiyuki Arima, Toyotaka Sonoda, Optimisation of a Stator Blade Used in a Transonic Compressor Cascade with Evolution Strategies Evolutionary Design and Manufacture. pp. 45- 54 ,(2000) , 10.1007/978-1-4471-0519-0_4
Christopher M. Bishop, Neural networks for pattern recognition ,(1995)
Mohammed A. El-Beltagy, Andy J. Keane, Prasanth B. Nair, Metamodeling techniques for evolutionary optimization of computationally expensive problems: promises and limitations genetic and evolutionary computation conference. pp. 196- 203 ,(1999)
Alain Ratle, Accelerating the convergence of evolutionary algorithms by fitness landscape approximation Lecture Notes in Computer Science. pp. 87- 96 ,(1998) , 10.1007/BFB0056852
David E. Goldberg, Genetic algorithms in search, optimization, and machine learning ,(1988)
J. Michael Fitzpatrick, John J. Grefenstette, Genetic Algorithms in Noisy Environments Machine Learning. ,vol. 3, pp. 101- 120 ,(1988) , 10.1007/BF00113893
Shigeru Obayashi, Yoshihiro Yamaguchi, Takashi Nakamura, Multiobjective Genetic Algorithm for Multidisciplinary Design of Transonic Wing Planform Journal of Aircraft. ,vol. 34, pp. 690- 693 ,(1997) , 10.2514/2.2231
Andreas Ostermeier, Andreas Gawelczyk, Nikolaus Hansen, A derandomized approach to self-adaptation of evolution strategies Evolutionary Computation. ,vol. 2, pp. 369- 380 ,(1994) , 10.1162/EVCO.1994.2.4.369
Thomas J Santner, Brian J Williams, William I Notz, Brain J Williams, None, The Design and Analysis of Computer Experiments ,(2003)