Olivier Teytaud
ABSTRACT
In a preliminary part of this paper, we analyze the necessity of randomness in evolutionstrategies. We conclude to the necessity of "continuous"-randomness, but with a much more limited use of randomness than whatis commonly used in evolution strategies. We then apply these results to CMA-ES, a famous evolution strategy already based on the idea of derandomization, which uses random independent Gaussian mutations. We here replace these random independent Gaussian mutations by a quasi-randomsample. The modification is very easy to do, the modified algorithm is computationally more efficient and its convergence is faster in terms of the number of iterates for a given precision.
- A. Auger, M. Jebalia, and O. Teytaud. Xse: quasi-random mutations for evolution strategies. In Proceedings of Evolutionary Algorithms, 12 pages, 2005. Google Scholar
Digital Library
- A. Auger, M. Schoenauer, and O. Teytaud. Local and global order 3/2 convergence of a surrogate evolutionary algorithm. In GECCO, pages 857--864, 2005. Google ScholarDigital Library
- H.-G. Beyer. The Theory of Evolutions Strategies. Springer, Heidelberg, 2001. Google ScholarDigital Library
- P. Bratley and B. Fox. Algorithm 659: Implementing sobol's quasirandom sequence generator. ACM Transactions on Mathematical Software Volume 14, Number 1, pages 88--100, 1988. Google ScholarDigital Library
- C. Cervellera and M. Muselli. A deterministic learning approach based on discrepancy. In Proceedings of WIRN'03, pp 53--60, 2003.Google ScholarCross Ref
- A. Conn, K. Scheinberg, and L. Toint. Recent progress in unconstrained nonlinear optimization without derivatives, 1997.Google Scholar
- R. Cranley and T. N. L. Patterson. Randomization of number theoretic methods for multiple integration. SIAM J. Numer. Anal, 13(6):904--914, Dec. 1976.Google ScholarCross Ref
- A. Eiben and J. Smith. Introduction to Evolutionary Computing. springer, 2003. Google ScholarDigital Library
- S. Gelly, J. Mary, and O. Teytaud. On the ultimate convergence rates for isotropic algorithms and the best choices among various forms of isotropy. In 10th International Conference on Parallel Problem Solving from Nature (PPSN 2006), 2006. Google ScholarDigital Library
- N. I. M. Gould, D. Orban, and P. L. Toint. Cuter and sifdec: A constrained and unconstrained testing environment, revisited. ACM Trans. Math. Softw., 29(4):373--394, 2003. Google ScholarDigital Library
- J. Halton. On the efficiency of certain quasi-random sequences of points in evaluating multi-dimensional integrals. Numerische Mathematik, 2:84--90, 1960.Google ScholarDigital Library
- N. Hansen and A. Ostermeier. Completely derandomized self-adaptation in evolution strategies. Evolutionary Computation, 11(1), 2003. Google ScholarDigital Library
- F. Hickernell. A generalized discrepancy and quadrature error bound. Math. Comp. 67, 299--322, 1998. Google ScholarDigital Library
- S. Kimura and K. Matsumura. Genetic algorithms using low-discrepancy sequences. In GECCO, pages 1341--1346, 2005. Google ScholarDigital Library
- P. L'Ecuyer and C. Lemieux. Recent advances in randomized quasi-monte carlo methods, 2002.Google Scholar
- H. Niederreiter. Random Number Generation and Quasi-Monte Carlo Methods. SIAM, 1992. Google ScholarDigital Library
- A. B. Owen. Quasi-Monte Carlo sampling. In H. W. Jensen, editor, Monte Carlo Ray Tracing: Siggraph 2003 Course 44, pages 69--88. SIGGRAPH, 2003.Google Scholar
- I. Rechenberg. Evolutionstrategie: Optimierung Technisher Systeme nach Prinzipien des Biologischen Evolution. Fromman-Hozlboog Verlag, Stuttgart, 1973.Google Scholar
- P. Sarkar and M. Prasad. A comparative study of pseudo and quasi random sequences for the solution of integral equations. J. Computational Physics, 68, pages 66--88, 1978. Google ScholarDigital Library
- I. Sloan and H. Wózniakowski. When are quasi-Monte Carlo algorithms efficient for high dimensional integrals? Journal of Complexity, 14(1):1--33, 1998. Google ScholarDigital Library
- I. M. Sobol. On the systematic search in a hypercube. j-SIAM-J-NUMER-ANAL, 16(5):790--793, Oct. 1979.Google Scholar
- P. N. Suganthan, N. Hansen, J. J. Liang, K. Deb, Y.-P. Chen, A. Auger, and S. Tiwari. Problem definitions and evaluation criteria for the cec 2005 special session on real-parameter optimization. Technical Report AND KanGAL Report #2005005, IIT Kanpur, India, Nanyang Technological University, Singapore, 2005.Google Scholar
- T.Back, F.Hoffmeister, and H.Schewefel. A survey of evolution strategies. Technical report, Dpt. of Computer Science XI, University of Dortmund, D-4600 , Dortmund 50, Germany, 1991.Google Scholar
- B. Tuffin. On the use of low discrepancy sequences in monte carlo methods. In Technical Report 1060, I.R.I.S.A., 1996.Google ScholarCross Ref
Index Terms
- DCMA: yet another derandomization in covariance-matrix-adaptation
Recommendations
Nondeterministic circuit lower bounds from mildly derandomizing Arthur-Merlin games
In several settings, derandomization is known to follow from circuit lower bounds that themselves are equivalent to the existence of pseudorandom generators. This leaves open the question whether derandomization implies the circuit lower bounds that are ...
QR Mutations Improve Many Evolution Strategies: A Lot On Highly Multimodal Problems
GECCO '16 Companion: Proceedings of the 2016 on Genetic and Evolutionary Computation Conference CompanionPrevious studies have shown the efficiency of using quasi- random mutations on the well-know CMA evolution strategy. Quasi-random mutations have many advantages, in particular their application is stable, efficient and easy to use. In this article, we ...
Mirrored orthogonal sampling with pairwise selection in evolution strategies
SAC '14: Proceedings of the 29th Annual ACM Symposium on Applied ComputingIn this paper, an improvement of the mirrored sampling method, called mirrored orthogonal sampling, is propsed. The convergence rates on the sphere function are estimated. It is also applied to Covariance Matrix Adaptation Evolution Strategy (CMA-ES). ...
Comments