Machine Learning for Evolution Strategies

Oliver Kramer

ISBN 10: 331933381X / ISBN 13: 9783319333816

Editore: Springer International Publishing, Springer International Publishing Jun 2016, 2016

Riguardo questo articolo

Neuware -This bookintroduces numerous algorithmic hybridizations between both worlds that showhow machine learning can improve and support evolution strategies. The set ofmethods comprises covariance matrix estimation, meta-modeling of fitness andconstraint functions, dimensionality reduction for search and visualization ofhigh-dimensional optimization processes, and clustering-based niching. Aftergiving an introduction to evolution strategies and machine learning, the bookbuilds the bridge between both worlds with an algorithmic and experimentalperspective. Experiments mostly employ a (1+1)-ES and are implemented in Pythonusing the machine learning library scikit-learn. The examples are conducted ontypical benchmark problems illustrating algorithmic concepts and theirexperimental behavior. The book closes with a discussion of related lines ofresearch.Springer Verlag GmbH, Tiergartenstr. 17, 69121 Heidelberg 136 pp. Englisch.

Codice articolo 9783319333816

Dati bibliografici

• Titolo: Machine Learning for Evolution Strategies
• Editore: Springer International Publishing, Springer International Publishing Jun 2016
• Data di pubblicazione: 2016
• Lingua: Inglese
• ISBN 10: 331933381X
• ISBN 13: 9783319333816
• Rilegatura: Buch
• Condizione: Neu
• Dimensioni: 241 L x 160 A x 14 P (millimetri) 241 millimetri di larghezza per 160 millimetri di altezza per 14 millimetri di profondità
• Peso: 377 grammi

Descrizione articolo

Riassunto

This bookintroduces numerous algorithmic hybridizations between both worlds that showhow machine learning can improve and support evolution strategies. The set ofmethods comprises covariance matrix estimation, meta-modeling of fitness andconstraint functions, dimensionality reduction for search and visualization ofhigh-dimensional optimization processes, and clustering-based niching. Aftergiving an introduction to evolution strategies and machine learning, the bookbuilds the bridge between both worlds with an algorithmic and experimentalperspective. Experiments mostly employ a (1+1)-ES and are implemented in Pythonusing the machine learning library scikit-learn. The examples are conducted ontypical benchmark problems illustrating algorithmic concepts and theirexperimental behavior. The book closes with a discussion of related lines ofresearch.