Honey is a very popular and yet it is one of the most mislabeled food item in the world. All over the world, it is becoming highly difficult to identify real honey. Even trusted suppliers tend to mix ingredients like sugar cane, rice syrups, and corn. Some suppliers also go to the extent of adding toxic elements like animal antibiotics, lead, and other heavy metals to it. All these are not safe for human intake. Therefore, it is crucial that real honey can be identified correctly.
The process of authenticating honey (Melissopalynology) from its botanical source is a costly and time-consuming process that is carried out in the lab by a specialist and with the use of specialized equipment. This process is too sophisticated hence the need for a better and simpler solution.
This is where machine learning comes in. Machine learning has the amazing ability to be able to classify things that are difficult for humans too. With machine learning, a more convenient and cheap method has been developed, and it is laid out in a paper titled “Honey Authentication with Machine Learning Augmented Bright-Field Microscopy.” This paper was accepted at the “AI for Social Good” workshop at last year’s Neural Information Processing Systems (NeurIPS), a prestigious AI conference.
This honey authentication tool was developed by two college students and a high-schooler using a $130 microscope. This microscope is said to be very easy to operate even for an 11- year old.
“We thus reckon that it would, in practice, prove scalable as a decentralized system where producers/consumers/beekeeping associations can test honey easily and help weed out fraudsters,” said Peter He, one of the authors of the paper and a student at Imperial College London, in an email.
During the process of developing the tool, they gathered different types of honey (manuka, acacia, “Lithuanian,” “Black Forest,” eucalyptus
The steps by which the machine learning works in identifying the authenticity of honey can be simplified into two:
1. The machine learning uses it’s pollen identification neural network to detect and determine the botanical origin, density, and distribution of present pollen,
2. It then uses this data to verify if the honey is what it says it is
“Honey samples diluted with sugar syrup can be detected from pollen density analysis, and honey samples diluted with cheaper kinds of honey can be detected from pollen distribution comparison,” the researchers wrote in the paper. “Mislabeled honey samples can be identified through the botanical sources of their pollen.”
So far the tool has not been able to identify contaminations with antibiotics, heavy metals or pesticides. But researchers believe this tool tends to make honey inspection easier, faster and cheaper. It is believed that with more research and funding they may be able to create honey fingerprints for specific producers. If these are achieved, it can be used in forensic palynology or the study of pollen and spores for legal matters.
Note: Some information used in this article are from https://www.fastcompany.com/90294831/machine-learning-is-a-sweet-way-to-tell-if-your-honey-is-fake