Data fusion - An effective tool for the development of recognition models for food authentication

The development of new reliable recognition models for food and beverage differentiation represents an assumed interest of many research groups. In this regard, data processing proved to have enormous importance in increasing the ability to extract the most relevant information, especially in the case when this is not straightforward. Data fusion strategies can be framed in the approaches that try to corroborate the outputs provided by two or more analytical techniques, with the general aim of improving the reliability of the prediction models. Against this background, our work proposes the association of the information provided by two complementary vibrational spectroscopy techniques, Attenuated Total Reflection Fourier Transform Infrared (ATR-FT-IR) and Fourier Transform Raman (FT-Raman) spectroscopy, for the development of new improved differentiation models for honey recognition. As working strategies, low- and mid-level data fusion methods were applied for our purpose. The recognition models constructed for the investigated samples aimed to discriminate honey with respect to the botanical origin and harvesting year.