An evaluation of different bio-inspired feature selection techniques on multivariate calibration models in spectroscopy.

Herein, two new swarm intelligence based algorithms namely; grey wolf optimization (GWO) and antlion optimization (ALO) algorithms were presented, for the first time, as variable selection tools in spectroscopic data analysis. In order to assess the performance of these algorithms, they were applied along with the recently introduced firefly algorithm (FFA) and the well-established genetic algorithm (GA) and particle swarm optimization (PSO) algorithm on four different spectroscopic datasets of varying sizes and nature (UV and IR). Partial least squares (PLS) regression models were built using the selected variables by these algorithms along with the full spectral data as the reference models. The obtained results prove that the ALO and GWO optimization algorithms select variables in most cases less than GA and PSO while keeping the PLS performance almost the same. Accordingly, these algorithms can be successfully used for variable selection in spectroscopic data analysis.