:: data-machine ::  
:: d a t a : m a c h i n e :: 
tools for computational intelligence
Global optimization: genetic algorithms
and a lot more






Neural networks

Fuzzy systems

Chaotic systems

and evolutionary

To receive
information about
the new updates
please send an email:

Many real-life problems require the solution of optimization problems. If the objective function is not known analytically, traditional gradient-based methods are not applicable and it is necessary to use direct methods that do not require calucation of derivatives. Typically we cannot assume existence of a single extremum, so multi-extremum (global) optimization (GO) methods must be used. One of the popular GO algorithms is a genetic algorithm (GA). Standard GAs appeared to be not very efficient, and other "poulation-based" algorithms have been also suggested: memetic algorithms, ant optimization, particle swarm optimization, and others. Older algorithms are used as well, for example, simplex downhill descent (Nelder and Mead 1965). Choice of an algorithm depends on its effectiveness (accuracy), efficiency (number of needed function evaluations) and reliability.

It has been shown that on many problems the adaptive cluster covering (ACCO) algorithm (Solomatine 1999) shows better performance than some of the popular algorithms, for example GA.

GLOBE tool incorporates nine algorithms and makes it possible to compare them. User can specify an objective function caluclated by an external EXEcutable or a DLL.

Download Global optimization tool GLOBE
NEW: Download Global Optimizer for MATLAB
GLOBE screen dumps (240K)

Some publications on using ACCO algorithm:

Solomatine, D.P. Two strategies of adaptive cluster covering with descent and their comparison to other algorithms. Journal of Global Optimization, 1999, vol. 14, No. 1, pp. 55-78.

Solomatine D.P. Genetic and other global optimization algorithms - comparison and use in model calibration. Proc. Intern Conf. Hydroinformatics-98, Balkema, Rotterdam, 1998.

D.P. Solomatine. Adaptive cluster covering and evolutionary approach: comparison, differences and similarities. Proc. IEEE Congress on Evolutionary Computation , Edinburgh , U.K. , 2005, 1959-1966.

Solomatine D.P., Dibike Y.B, Kukric N. Automatic Calibration of Groundwater Models: Using Global Optimization Techniques. Hydrological Sciences J. 1999, 44(6), 1999, 879-894 .

Abebe A.J., Solomatine D.P. Application of global optimization to the design of pipe networks. Proc. Intern Conf. Hydroinformatics-98, Balkema, Rotterdam, 1998.

Other resources on optimization:

Neumaier's Global Optimization page
Decision Tree for Optimization Software
Glasgow Genetic Algorithm Demonstrator