Abstract
To parameterize continuous functions for evolutionary learning, we use kernel expansions in nested sequences of function spaces of growing complexity. This approach is particularly powerful when dealing with non-convex constraints and discontinuous objective functions. Kernel methods offer a number of beneficial properties for parameterizing continuous functions, such as smoothness and locality, which make them attractive as a basis for mutation operators. Beyond such practical considerations, kernel methods make heavy use of inner products in function space and offer a well established regularization framework. We show how evolutionary computation can profit from these properties. Searching function spaces of iteratively increasing complexity allows the solution to evolve from a simple first guess to a complex and highly refined function. At transition points where the evolution strategy is confronted with the next level of functional complexity, the kernel framework can be used to project the search distribution into the extended search space. The feasibility of the method is demonstrated on challenging trajectory planning problems where redundant robots have to avoid obstacles.
Similar content being viewed by others
Notes
Care has to be taken that the function space is restricted to L 2 functions, that is, functions of finite norm.
For any (not necessarily square) matrix M we write M −1 for the Moore–Penrose pseudo-inverse.
References
Alfaro T, Rojas MCR (2005) An on-the-fly evolutionary algorithm for robot motion planning. In: ICES, pp 119–130
Barraquand J, Latombe JC (1991) Robot motion planning: a distributed representation approach. Int J Robot Res 10(6):628–649
Beyer HG, Schwefel HP (2002) “Evolution strategies”—a comprehensive introduction. Nat Comput 1:3–52
Boser BE, Guyon IM, Vapnik VN (1992) A training algorithm for optimal margin classifiers. In: Proceedings of the fifth annual workshop on computational learning theory. COLT ’92, ACM, New York, NY, USA, pp 144–152
Conkur ES, Buckingham R (1997) Manoeuvring highly redundant manipulators. Robotica 15:435–447
Cortes C, Vapnik V (1995) Support-vector networks. Mach Learn 20(3):273–297
Denker A, Atherton D (1994) No-overshoot control of robotic manipulators in the presence of obstacles. J Robot Syst 11(7):665–678
Eiben AE, Smith JE (2003) Introduction to evolutionary computing. Springer, Berlin
Floreano D, Mitri S, Perez-Uribe A, Keller L (2008) Evolution of altruistic robots. In: Proceedings of the WCCI 2008, vol 5050. Springer, Berlin, pp 232–248
Glasmachers T, Schaul T, Schmidhuber J (2010) A natural evolution strategy for multi-objective optimization. In: Parallel problem solving from nature (PPSN)
Glasmachers T, Schaul T, Sun Y, Wierstra D, Schmidhuber J (2010) Exponential natural evolution strategies. In: Genetic and evolutionary computation conference (GECCO), Portland, OR
Gomez F, Schmidhuber J, Miikkulainen R (2006) Efficient non-linear control through neuroevolution. In: Fürnkranz J, Scheffer T, Spiliopoulou M (eds) Proceeding of the European conference on machine learning, No. 4212 in LNAI, Springer, pp 654–662
Hansen N, Ostermeier A (2001) Completely derandomized self-adaptation in evolution strategies. Evol Comput 9(2):159–195
Harding S, Miller JF (2005) Evolution of robot controller using Cartesian genetic programming. Genetic programming, pp 62–73
Hayashi A (1994) Geometrical motion planning for highly redundant manipulators using a continuous model. PhD thesis, University of Texas Austin
Iossifidis I, Schöner G (2006) Dynamical systems approach for the autonomous avoidance of obstacles and joint-limits for an redundant robot arm. In: IEEE/RSJ international conference on intelligent robots and systems, pp 580–585
Kavraki L, Svestka P, Latombe JC, Overmars M (1996) Probabilistic roadmaps for path planning in high-dimensional configuration spaces. In: IEEE international conference on robotics and automation, pp. 566–580
Khatib O (1986) Real-time obstacle avoidance for manipulators and mobile robots. Int J Robot Res 5(1):90–98
Koutník J, Gomez FJ, Schmidhuber J (2010) Evolving neural networks in compressed weight space. In: GECCO, pp 619–626
Latombe JC (1991) Robot motion planning. Kluwer, Norwell
Lee JD, Wang BL (1988) Optimal control of a flexible robot arm. Comput Struct 29(3):459–467
Mitrovic D, Klanke S, Vijayakumar S (2010) Adaptive optimal feedback control with learned internal dynamics models. In: Sigaud O, Peters J (eds) From motor learning to interaction learning in robots. Springer, Berlin, pp 65–84
Nolfi S, Marocco D (2001) Evolving robots able to integrate sensory-motor information over time. Theory Biosci 120:287–310
Rechenberg I, Eigen M (1973) Evolutionsstrategie: Optimierung technischer Systeme nach Prinzipien der biologischen Evolution. Frommann-Holzboog, Stuttgart
Schaul T, Glasmachers T, Schmidhuber J (2011) High dimensions and heavy tails for natural evolution strategies. In: Proceedings of the genetic and evolutionary computation conference (GECCO)
Scholkopf B, Smola A, Muller KR (1998) Nonlinear component analysis as a Kernel eigenvalue problem. Neural Comput 10(5):1299–1319
Sun Y, Wierstra D, Schaul T, Schmidhuber J (2009) Efficient natural evolution strategies. In: Genetic and evolutionary computation conference (GECCO)
Sun Y, Wierstra D, Schaul T, Schmidhuber J (2009) Stochastic search using the natural gradient. In: International conference on machine learning (ICML)
Vapnik V (1997) The support vector method. In: ICANN, pp 263–271
Vapnik VN (1998) Statistical learning theory. Wiley-Interscience, Hoboken
Wierstra D, Schaul T, Peters J, Schmidhuber J (2008) Natural evolution strategies. In: Proceedings of the congress on evolutionary computation (CEC08), Hongkong. IEEE Press
Woolley BG, Stanley KO (2010) Evolving a single scalable controller for an octopus arm with a variable number of segments. In: PPSN (2), pp 270–279
Yekutieli Y, Sagiv-Zohar R, Aharonov R, Engel Y, Hochner B, Flash T (2005) A dynamic model of the octopus arm. I. Biomechanics of the octopus reaching movement. J Neurophysiol 94(2):1443–1458
Acknowledgments
This work was funded through the 7th Framework Programme of the EU under grant number 231576 (STIFF project) and SNF grant 200020-125038/1.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Glasmachers, T., Koutník, J. & Schmidhuber, J. Kernel representations for evolving continuous functions. Evol. Intel. 5, 171–187 (2012). https://doi.org/10.1007/s12065-012-0070-y
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s12065-012-0070-y