Efficient natural evolution strategies

Authors:
Yi Sun;Daan Wierstra;Tom Schaul;Juergen Schmidhuber
Affiliations:
IDSIA, Manno, Switzerland;IDSIA, Manno, Switzerland;IDSIA, Manno, Switzerland;IDSIA, Manno, Switzerland
Venue:
Proceedings of the 11th Annual conference on Genetic and evolutionary computation
Year:
2009

Citing 5
Cited 9

Natural gradient works efficiently in learning

Neural Computation
Completely Derandomized Self-Adaptation in Evolution Strategies

Evolutionary Computation
Machine learning of motor skills for robotics

Machine learning of motor skills for robotics
Natural Actor-Critic

Neurocomputing
Stochastic search using the natural gradient

ICML '09 Proceedings of the 26th Annual International Conference on Machine Learning

Stochastic search using the natural gradient

ICML '09 Proceedings of the 26th Annual International Conference on Machine Learning
PyBrain

The Journal of Machine Learning Research
Exponential natural evolution strategies

Proceedings of the 12th annual conference on Genetic and evolutionary computation
Stochastic local search in continuous domains: questions to be answered when designing a novel algorithm

Proceedings of the 12th annual conference companion on Genetic and evolutionary computation
RGP: an open source genetic programming system for the R environment

Proceedings of the 12th annual conference companion on Genetic and evolutionary computation
Bidirectional relation between CMA evolution strategies and natural evolution strategies

PPSN'10 Proceedings of the 11th international conference on Parallel problem solving from nature: Part I
A natural evolution strategy for multi-objective optimization

PPSN'10 Proceedings of the 11th international conference on Parallel problem solving from nature: Part I
High dimensions and heavy tails for natural evolution strategies

Proceedings of the 13th annual conference on Genetic and evolutionary computation
Scalable neuroevolution for reinforcement learning

TPNC'12 Proceedings of the First international conference on Theory and Practice of Natural Computing

Quantified Score

Hi-index	0.00

Visualization

Abstract

Efficient Natural Evolution Strategies (eNES) is a novel alternative to conventional evolutionary algorithms, using the natural gradient to adapt the mutation distribution. Unlike previous methods based on natural gradients, eNES uses a fast algorithm to calculate the inverse of the exact Fisher information matrix, thus increasing both robustness and performance of its evolution gradient estimation, even in higher dimensions. Additional novel aspects of eNES include optimal fitness baselines and importance mixing (a procedure for updating the population with very few fitness evaluations). The algorithm yields competitive results on both unimodal and multimodal benchmarks.