Genetic algorithms incorporating a pseudo-subspace method

摘要: GA performance in high-dimensional optimisation problems can be enhanced by the use of a 'pseudo subspace' technique. The method works projecting parameter space onto lower dimensional subspace first stages process, order to allow search discover most promising area solution space. Subsequently, dimensionality model is progressively increased until predetermined limit reached. Comparison between pseudo-subspace procedure and conventional GA, using two different implementations, shows former more successful when applied geophysical characterised solution-space geometry mathematics. This technique could easily transferred image processing or pattern recognition where geometrical relationships parameters are maintained.