Estimation of distribution algorithm: Difference between revisions

Revision as of 01:17, 30 October 2016

Revision as of 04:26, 25 December 2016

Line 143:

===[[Bayesian optimization algorithm]] (BOA)===

The BOA<ref>{{cite journal|last1=Pelikan|first1=Martin|last2=Goldberg|first2=David E.|last3=Cantu-Paz|first3=Erick|title=BOA: The Bayesian Optimization Algorithm|date=1 January 1999|pages=525–532|url=http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.46.8131|publisher=Morgan Kaufmann}}</ref><ref>{{cite book|last1=Pelikan|first1=Martin|title=Hierarchical Bayesian optimization algorithm : toward a new generation of evolutionary algorithms|date=2005|publisher=Springer|___location=Berlin [u.a.]|isbn=978-3-540-23774-7|edition=1st ed.}}</ref><ref>{{cite journal|last1=Wolpert|first1=David H.|last2=Rajnarayan|first2=Dev|title=Using Machine Learning to Improve Stochastic Optimization|journal=Proceedings of the 17th AAAI Conference on Late-Breaking Developments in the Field of Artificial Intelligence|date=1 January 2013|pages=146–148|url=http://dl.acm.org/citation.cfm?id=2908286.2908335|publisher=AAAI Press}}</ref> uses Bayesian networks to model and sample promising solutions. Bayesian networks are directed acyclic graphs, with nodes representing variables and edges representing conditional probabilities between pair of variables. The value of a variable <math>x_i</math> can be conditioned on a maximum of <math>K</math> other variables, defined in <math>\pi_i</math>. BOA builds a PGM encoding a factorized joint distribution, in which the parameters of the network, i.e. the conditional probabilities, are estimated from the selected population using the maximum likelihood estimator.

<math>