Abstract
Recently, we applied successfully ant colony optimization (ACO) to wavelength selection problem. In this paper ant colony system (ACS), a version of ACO algorithms, has been applied to variable selection in multivariate calibration and QSPR studies. In addition, the effect of combination of ACS algorithm with various local search strategies was inspected. The suggested local search strategies included forward selection, backward elimination, forward selection together with backward elimination, backward elimination together with forward selection, genetic algorithm, and tabu search. Parameters of ACS and combined local search strategy were optimized. The results obtained from all ACS algorithms indicated that combination of various local search strategies to the ACS algorithm improved the performance of the ACS algorithm