Russian scientists have trained AI to identify counterfeit milk and its components

Brief retelling from RIA II

Russian scientists have trained artificial intelligence to identify various types of falsified milk, as well as find antibiotics, preservatives, herbal supplements and genetically modified components in it.

Now experts use different methods to assess the quality of milk, but none of them completely solves the problem of falsification.

The researchers proposed building the inspection of dairy products according to a multi-level scheme: first, rapid non-destructive screening, then, if falsification is suspected, precise laboratory verification.

MOSCOW, September 23 - RIA Novosti. Russian scientists have trained artificial intelligence to identify various types of falsified milk, as well as find antibiotics, preservatives, herbal additives and genetically modified components in it, Kazan State Agrarian University told RIA Novosti.

Nowadays there are different forms of adulterated milk on the market. “Part of the milk fat can be replaced with vegetable fats, whole milk is diluted with water or skim milk, the milk of one animal is passed off as the milk of another, and sometimes the information on the packaging or in accompanying documents is simply distorted,” the researchers said.

To assess the quality of milk, experts now use different methods, but none of them by itself completely solves the problem of falsification.

For example, physicochemical methods allow you to measure the fat and protein content in milk, chromatographic and spectroscopic methods help you find antibiotics, preservatives and the presence of added water. And molecular biological methods make it possible to determine the species of milk, find vegetable protein or genetically modified components in it.

All these methods provide a huge amount of compositional data for each sample. “It is at this stage that machine learning algorithms perform more accurately than classical statistical calculations: they simultaneously take into account dozens of spectrum parameters and find patterns that are difficult for humans to notice,” the scientists noted.