Advertisement

On Low Complexity Acceleration Techniques for Randomized Optimization

  • Sebastian Urban Stich
Part of the Lecture Notes in Computer Science book series (LNCS, volume 8672)

Abstract

Recently it was shown by Nesterov (2011) that techniques form convex optimization can be used to successfully accelerate simple derivative-free randomized optimization methods. The appeal of those schemes lies in their low complexity, which is only Θ(n) per iteration—compared to Θ(n 2) for algorithms storing second-order information or covariance matrices. From a high-level point of view, those accelerated schemes employ correlations between successive iterates—a concept looking similar to the evolution path used in Covariance Matrix Adaptation Evolution Strategies (CMA-ES). In this contribution, we (i) implement and empirically test a simple accelerated random search scheme (SARP). Our study is the first to provide numerical evidence that SARP can effectively be implemented with adaptive step size control and does not require access to gradient or advanced line search oracles. We (ii) try to empirically verify the supposed analogy between the evolution path and SARP. We propose an algorithm CMA-EP that uses only the evolution path to bias the search. This algorithm can be generalized to a family of low memory schemes, with complexity Θ(mn) per iteration, following a recent approach by Loshchilov (2014). The study shows that the performance of CMA-EP heavily depends on the spectra of the objective function and thus it cannot accelerate as consistently as SARP.

Keywords

Gradient-free optimization accelerated random search evolution path adaptive step size Covariance Matrix Adaptation spectra 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    Polyak, B.: Introduction to Optimization. Optimization Software - Inc. (1987)Google Scholar
  2. 2.
    Nesterov, Y.: Introductory Lectures on Convex Optimization. Kluwer (2004)Google Scholar
  3. 3.
    Broyden, C.G.: The Convergence of a Class of Double-rank Minimization Algorithms 1. General Considerations. IMA J. of Appl. Math. 6(1), 76–90 (1970)MathSciNetCrossRefzbMATHGoogle Scholar
  4. 4.
    Fletcher, R.: A new approach to variable metric algorithms. The Computer Journal 13(3), 317–322 (1970)MathSciNetCrossRefzbMATHGoogle Scholar
  5. 5.
    Goldfarb, D.: A Family of Variable-Metric Methods Derived by Variational Means. Mathematics of Computation 24(109), 23–26 (1970)MathSciNetCrossRefzbMATHGoogle Scholar
  6. 6.
    Nocedal, J.: Updating Quasi-Newton Matrices with Limited Storage. Mathematics of Computation 35(151), 773–782 (1980)MathSciNetCrossRefzbMATHGoogle Scholar
  7. 7.
    Liu, D.C., Nocedal, J.: On the limited memory BFGS method for large scale optimization. Mathematical Programming 45(1-3), 503–528 (1989)MathSciNetCrossRefzbMATHGoogle Scholar
  8. 8.
    Nesterov, Y.: A method of solving a convex programming problem with convergence rate O(1/k 2). Soviet Mathematics Doklady 27(2), 372–376 (1983)zbMATHGoogle Scholar
  9. 9.
    Nesterov, Y.: Smoothing technique and its applications in semidefinite optimization. Mathematical Programming 110(2), 245–259 (2007)MathSciNetCrossRefzbMATHGoogle Scholar
  10. 10.
    Tseng, P.: On accelerated proximal gradient methods for convex-concave optimization. Submitted to SIAM Journal on Optimization (2008)Google Scholar
  11. 11.
    Schumer, M., Steiglitz, K.: Adaptive step size random search. IEEE Transactions on Automatic Control 13(3), 270–276 (1968)CrossRefGoogle Scholar
  12. 12.
    Rechenberg, I.: Evolutionsstrategie; Optimierung technischer Systeme nach Prinzipien der biologischen Evolution. Frommann-Holzboog (1973)Google Scholar
  13. 13.
    Mutseniyeks, V.A., Rastrigin, L.A.: Extremal control of continuous multi-parameter systems by the method of random search. Eng.Cyb. 1, 82–90 (1964)Google Scholar
  14. 14.
    Stich, S.U., Müller, C.L., Gärtner, B.: Optimization of convex functions with Random Pursuit. SIAM Journal on Optimization 23(2), 1284–1309 (2013)MathSciNetCrossRefzbMATHGoogle Scholar
  15. 15.
    Nesterov, Y.: Random Gradient-Free Minimization of Convex Functions. Technical report, ECORE (2011)Google Scholar
  16. 16.
    Leventhal, D., Lewis, A.S.: Randomized Hessian estimation and directional search. Optimization 60(3), 329–345 (2011)MathSciNetCrossRefzbMATHGoogle Scholar
  17. 17.
    Stich, S.U., Gärtner, B., Müller, C.L.: Variable Metric Random Pursuit (2012) (submitted), http://arxiv.org/abs/1210.5114
  18. 18.
    Hansen, N., Ostermeier, A.: Completely Derandomized Self-Adaption in Evolution Strategies. Evolutionary Computation 9(2), 159–195 (2001)CrossRefGoogle Scholar
  19. 19.
    Hansen, N., Muller, S.D., Koumoutsakos, P.: Reducing the time complexity of the derandomized evolution strategy with covariance matrix adaptation (CMA-ES). Evol. Comput. 11(1), 1–18 (2003)CrossRefGoogle Scholar
  20. 20.
    Knight, J.N., Lunacek, M.: Reducing the Space-time Complexity of the CMA-ES. In: GECCO 2007, pp. 658–665. ACM (2007)Google Scholar
  21. 21.
    Loshchilov, I.: A Computationally Efficient Limited Memory CMA-ES for Large Scale Optimization. To appear GECCO (2014), http://arxiv.org/abs/1404.5520
  22. 22.
    Lee, Y.T., Sidford, A.: Efficient accelerated coordinate descent methods and faster algorithms for solving linear systems. In: FOCS, pp. 147–156. IEEE (2013) Google Scholar
  23. 23.
    Ostermeier, A., Gawelczyk, A., Hansen, N.: Step-size adaptation based on non-local use of selection information. In: Davidor, Y., Männer, R., Schwefel, H.-P. (eds.) PPSN 1994. LNCS, vol. 866, pp. 189–198. Springer, Heidelberg (1994)CrossRefGoogle Scholar
  24. 24.
    Igel, C., Suttorp, T., Hansen, N.: A Computational Efficient Covariance Matrix Update and a (1+1)-CMA for Evolution Strategies. In: GECCO, pp. 453–460 (2006)Google Scholar
  25. 25.
    Sun, Y., Schaul, T., Gomez, F., Schmidhuber, J.: A linear time natural evolution strategy for non-separable functions. In: Proc. 15th Genetic and Evolutionary Computation Conference Companion, pp. 61–62. ACM (2013)Google Scholar
  26. 26.
    Stich, S.U.: Supplementary Online Mat (2014), http://arxiv.org/abs/1406.2010
  27. 27.
    Stich, S.U., Müller, C.L.: On Spectral Invariance of Randomized Hessian and Covariance Matrix Adaptation Schemes. In: Coello, C.A.C., Cutello, V., Deb, K., Forrest, S., Nicosia, G., Pavone, M. (eds.) PPSN 2012, Part I. LNCS, vol. 7491, pp. 448–457. Springer, Heidelberg (2012)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  • Sebastian Urban Stich
    • 1
  1. 1.Institute of Theoretical Computer ScienceETH ZürichZürichSwitzerland

Personalised recommendations