META-EVOLUTIONARY OPTIMIZATION
Ales Gottvald
Institute of Scientific Instruments, Academy of Sciences of the CR,
Kralovopolska 147, CZ-612 64 Brno, Czech Republic; E-mail: gott@isibrno.cz
Abstract: Evolution Strategy (ES) is a stochastic optimization method inspired by a biological evolution theory. It shares many attributes with some other stochastic optimization methods (Simulated Annealing, Genetic Algorithms). Meta-optimization means optimizing optimization itself, and Meta-Evolution is its special case. It is a fundamental concept behind a general methodology developed for solving a large class of inverse and optimization problems in many applications. Meta-optimization brings some new qualities into the stochastic optimization: (i) enforced global convergence; (ii) reduced sensitivity with respect to initial configuration; (iii) optimal values of some auxiliary optimization parameters; (iv) unified implementation of multistart, relaxation and other global algorithms; (v) construction of total accuracy estimates and fundamental statistical limits. Are mortality and variability of individual species only statistical consequences of Meta-Evolutionary optimizations?
Back to: