A hierarchy machine: learning to optimize from nature and humans

  • Authors:
  • Martin Pelikan;David E. Goldberg

  • Affiliations:
  • Illinois Genetic Algorithms Laboratory (IlliGAL), Department of General Engineering, University of Illinois at Urbana-Champaign, 104 S. Mathews Avenue, Urbana, Illinois;Illinois Genetic Algorithms Laboratory (IlliGAL), Department of General Engineering, University of Illinois at Urbana-Champaign, 104 S. Mathews Avenue, Urbana, Illinois

  • Venue:
  • Complexity
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

This article proposes a competent hierarchical optimization method called the hierarchical Bayesian optimization algorithm (hBOA). hBOA extends the Bayesian optimization algorithm (BOA) by incorporating three important features for robust and scalable optimization of hierarchical problems: proper decomposition, chunking, and preservation of alternative solutions. Additionally, the article proposes a class of difficult hierarchical problems called hierarchical traps, hBOA is shown to provide a scalable solution to the class of hierarchically decomposable problems and anything easier. Specifically, hBOA can solve hierarchical traps and other nearly decomposable problems in approximately O(n1.55 log n) to O(n2) function evaluations, where n is the number of decision variables in the problem.