Genetic profiling of olives for location of origin and variety discrimination using Machine Learning

Abstract

Genetic profiling via biomarkers in the food industry is a technology that gains momentum in the context of quality assurance and protection against fraud, as well as securing commercial assets like designation of origin. However, current solutions are based on methods that require significant computational resources and management of large data volumes, making them unsuitable for applications in the context of Internet-of-Things (IoT), edge computing and microcontrollers (MCU). This study presents a novel, computationally efficient and robust approach for fully field-integrated, low-complexity and high-accuracy classification of olives variety and location of origin, based on genetic ‘fingerprinting’ via a minimal set of information-rich features. The method is tested with real-world datasets, achieving accuracy rates above 96% and 99%, respectively, using various instance-based and tree ensemble classification models.