Naive Bayesian classifier detecting phenethylamines based on their vibrational spectra and associated eigenvalues

Abstract

In this paper we are presenting an artificial intelligence application designed to screen for controlled phenethylamines. The training set consists of 30 vibrational spectra of stimulant and hallucinogenic amphetamines, as well as of negatives (non-amphetamines representing randomly selected chemicals of forensic interest). The Naive Bayesian Classifier indicates the likelihood that a substance belongs to one of the predefined classes. For this aim, in our case, the Principal Component Analysis (PCA) scores of the targeted compounds are subjected to a Naive Bayesian Classifier. The advantages of combining these two pattern recognition methods over the use of each method independently are discussed from the point of view of the detection efficiency.