Illinois Genetic Algorithms Laboratory

Pelikan, M., Goldberg, D.E., Ocenasek, J., Trebst, S. (2003)
TR No.: 2003019 | Download PDF | Download PS

Abstract:
One of the most important challenges in computational optimization is the design of advanced black-box optimization techniques that would enable automated, robust, and scalable solution to challenging optimization problems. This paper presents an advanced black-box optimizer—the hierarchical Bayesian optimization algorithm (hBOA)—that combines techniques of genetic and evolutionary computation, machine learning, and statistics to create a widely applicable tool for solving real-world optimization problems. The paper motivates hBOA, describes its basic procedure, and provides an in-depth empirical analysis of hBOA on the class of random 2D and 3D Ising spin glass problems. By deriving inspiration from natural evolution and the way the best problem solvers—humans—solve their problems, hBOA becomes capable of solving challenging real-world problems quickly, accurately, and reliably. The
results on Ising spin glasses indicate that even without the need for
prior problem-specific knowledge hBOA can sometimes beat specialized
algorithms that exploit all available knowledge about the problem to speed
up the search as much as possible. Furthermore, hBOA can solve a large
class of hierarchically decomposable problems intractable by any other
algorithm.