Spectral quantitative analysis of complex samples based on the extreme learning machine

Multivariate calibrations, including linear and non-linear methods, have been widely used in the spectral quantitative analysis of complex samples. Despite their efficiency and few parameters involved, linear methods are inferior for nonlinear problems. Non-linear methods also have disadvantages such as the requirement of many parameters, time-consuming and easily relapses into local optima though the outstanding performance in nonlinearity. Thus, taking the advantages of both linear and non-linear methods, a novel algorithm called the extreme learning machine (ELM) is introduced. The efficiency and stability of this method are investigated first. Then, the optimal activation function and number of hidden layer nodes are determined by a newly defined parameter, which takes into account both the predictive accuracy and stability of the model. The predictive performance of ELM is compared with principal component regression (PCR), partial least squares (PLS), support vector regression (SVR) and back propagation artificial neural network (BP-ANN) by three near-infrared (NIR) spectral datasets of diesel fuel, a ternary mixture and blood. Results show that the efficiency of ELM is mainly affected by the number of nodes for a certain dataset. Despite some instability, ELM becomes stable close to the optimal parameters. Moreover, ELM has a better or comparable performance compared with its competitors in the spectral quantitative analysis of complex samples.