A Frequency-Domain Encoding for Neuroevolution

Neuroevolution has yet to scale up to complex reinforcement learning tasks that require large networks. Networks with many inputs (e.g. raw video) imply a very high dimensional search space if encoded directly. Indirect methods use a more compact genotype representation that is transformed into networks of potentially arbitrary size. In this paper, we present an indirect method where networks are encoded by a set of Fourier coefficients which are transformed into network weight matrices via an inverse Fourier-type transform. Because there often exist network solutions whose weight matrices contain regularity (i.e. adjacent weights are correlated), the number of coefficients required to represent these networks in the frequency domain is much smaller than the number of weights (in the same way that natural images can be compressed by ignore high-frequency components). This "compressed" encoding is compared to the direct approach where search is conducted in the weight space on the high-dimensional octopus arm task. The results show that representing networks in the frequency domain can reduce the search-space dimensionality by as much as two orders of magnitude, both accelerating convergence and yielding more general solutions.

[1]  Risto Miikkulainen,et al.  Evolving Neural Networks through Augmenting Topologies , 2002, Evolutionary Computation.

[2]  Tom Schaul,et al.  A linear time natural evolution strategy for non-separable functions , 2011, GECCO.

[3]  Hiroaki Kitano,et al.  Designing Neural Networks Using Genetic Algorithms with Graph Generation System , 1990, Complex Syst..

[4]  Kenneth O. Stanley,et al.  Generating large-scale neural networks through discovering geometric regularities , 2007, GECCO '07.

[5]  Jürgen Schmidhuber,et al.  Searching for Minimal Neural Networks in Fourier Space , 2010, AGI 2010.

[6]  Corso Elvezia Discovering Solutions with Low Kolmogorov Complexity and High Generalization Capability , 1995 .

[7]  Jan Koutník,et al.  NEAT in HyperNEAT Substituted with Genetic Programming , 2009, ICANNGA.

[8]  Jürgen Schmidhuber,et al.  Evolving neural networks in compressed weight space , 2010, GECCO '10.

[9]  Risto Miikkulainen,et al.  Accelerated Neural Evolution through Cooperatively Coevolved Synapses , 2008, J. Mach. Learn. Res..

[10]  Ming Li,et al.  An Introduction to Kolmogorov Complexity and Its Applications , 2019, Texts in Computer Science.

[11]  Jürgen Schmidhuber,et al.  Discovering Neural Nets with Low Kolmogorov Complexity and High Generalization Capability , 1997, Neural Networks.

[12]  Frank Kirchner,et al.  A common genetic encoding for both direct and indirect encodings of networks , 2007, GECCO '07.

[13]  Tamar Flash,et al.  Dynamic model of the octopus arm. I. Biomechanics of the octopus reaching movement. , 2005, Journal of neurophysiology.

[14]  Tom Schaul,et al.  Towards Practical Universal Search , 2010, AGI 2010.

[15]  Kenneth O. Stanley,et al.  Evolving a Single Scalable Controller for an Octopus Arm with a Variable Number of Segments , 2010, PPSN.

[16]  Jürgen Schmidhuber,et al.  Kernel representations for evolving continuous functions , 2012, Evol. Intell..

[17]  Tom Schaul,et al.  High dimensions and heavy tails for natural evolution strategies , 2011, GECCO '11.

[18]  Tom Schaul,et al.  Natural Evolution Strategies , 2008, 2008 IEEE Congress on Evolutionary Computation (IEEE World Congress on Computational Intelligence).