Released 2013
Markovian Learning Estimation of Distribution Algorithm (MARLEDA) is an Estimation of Distribution Algorithm (EDA) that employs a Markov Random Field model. EDAs in general combine genetic algorithms with statistical modeling in order to learn and exploit the structure of search domains. Most EDAs use directed acyclic graphs (DAGs) as models; while they are useful in many areas, DAGs have inherent restrictions that make undirected graph models a viable alternative in some domains. Markov Random Fields allows constructing such an undirected model. Example tasks (in this package) include OneMax, deceptive trap functions, the 2D Rosenbrock function, and 2D Ising spin glasses. Potential applications include design of autonomous agents as well as optimization in computational biology, such as RNA structure prediction (described e.g. in Alden's PhD thesis ).
Matthew Alden Ph.D. Alumni mealden [at] uw edu
Risto Miikkulainen Faculty risto [at] cs utexas edu
MARLEDA: Effective Distribution Estimation through Markov Random Fields 2016
Matthew Alden and Risto Miikkulainen, Theoretical Computer Science, Vol. 633 (2016), pp. 4-18.
MARLEDA: Effective Distribution Estimation Through Markov Random Fields 2007
Matthew Alden, PhD Thesis, Department of Computer Sciences, the University of Texas at Austin. Also Technical Report AI07-349.