Discrimination of adulteration cow milk by improved υ-support vector machines (υ-SVM) and near infrared spectroscopy

Raw cow milk may be adulterated water and pseudo proteins such as melamine either intentionally or accidentally for getting more profit. This study prepared a series of adulteration milk samples by adding different amounts of aqueous solutions of dextrin (or starch) containing pseudo proteins (melamine, urea, or ammonium nitrate) to raw cow milks, which were collected from pastures surrounding Shanghai in 2011. Improved υ-support vector machines (υ-SVM) method was applied to build classification models for discriminating adulteration milks based on near infrared (NIR) spectra of different sample sets. Uniform design table was used to find good value of parameters of υ and σ. Results show that the adulteration milks in which content of the adulteration solutions equals or exceeds 5% can be well identified by improved υ-SVM combining NIR spectroscopy with high validation correct ratio (VCR), which positively correlates to the content of the adulteration solutions. Water amount added to raw milks is the key fact influencing classification results of the NIR classification models. However, it is difficult to judge type of pseudo proteins added to milks by NIR spectroscopy. Chromatography or kit method aiming to particular prohibited substances such as melamine could be selected to further detect which kind of pseudo proteins is added to milk if necessary.