Neuroevolution strategies for episodic reinforcement learning

Created by W.Langdon from gp-bibliography.bib Revision:1.3973

@Article{HeidrichMeisner2009152,
  author =       "Verena Heidrich-Meisner and Christian Igel",
  title =        "Neuroevolution strategies for episodic reinforcement
                 learning",
  journal =      "Journal of Algorithms",
  volume =       "64",
  number =       "4",
  pages =        "152--168",
  year =         "2009",
  note =         "Special Issue: Reinforcement Learning",
  ISSN =         "0196-6774",
  DOI =          "doi:10.1016/j.jalgor.2009.04.002",
  URL =          "http://www.sciencedirect.com/science/article/B6WH3-4W7RY8J-3/2/22f7075bc25dab10a8ff3714e2fee303",
  keywords =     "genetic algorithms, genetic programming, Reinforcement
                 learning, Evolution strategy, Covariance matrix
                 adaptation, Partially observable Markov decision
                 process, Direct policy search",
  abstract =     "Because of their convincing performance, there is a
                 growing interest in using evolutionary algorithms for
                 reinforcement learning. We propose learning of neural
                 network policies by the covariance matrix adaptation
                 evolution strategy (CMA-ES), a randomised
                 variable-metric search algorithm for continuous
                 optimisation. We argue that this approach, which we
                 refer to as CMA Neuroevolution Strategy (CMA-NeuroES),
                 is ideally suited for reinforcement learning, in
                 particular because it is based on ranking policies (and
                 therefore robust against noise), efficiently detects
                 correlations between parameters, and infers a search
                 direction from scalar reinforcement signals. We
                 evaluate the CMA-NeuroES on five different (Markovian
                 and non-Markovian) variants of the common pole
                 balancing problem. The results are compared to those
                 described in a recent study covering several RL
                 algorithms, and the CMA-NeuroES shows the overall best
                 performance.",
  notes =        "compared against CE \cite{gruau:1996:ceVdeGNN}",
}

Genetic Programming entries for Verena Heidrich-Meisner Christian Igel

Citations