Classification of Honey as Genuine or Fake via Artificial Neural Network using Gradient Descent Backpropagation Algorithm

Abstract

Honey is always among the lists for food fraud around the world. A whistleblower surfaced in a South African Honey manufacturer that claims their honey is revealed as passing off a sugar concoction. Also, in the Philippines, mid-2016 there's a manufacturer of honey named Cem's Honey that mislabels their product as real honey but in fact their product is a fake honey. The main objective of this study is to create a system that could classify a honey whether it is genuine or fake using Artificial Neural Network with Gradient Descent Backpropagation as the training algorithm, and sensors (Electrical Conductivity and pH Sensor). Through testing, the system classified the presented samples at an accuracy rate of 87.5%. In conclusion, the researchers successfully developed a system that can classify a honey whether is it genuine or fake using Artificial Neural Network.