Train AI to 'smell' wine for traceability

Professor Alexandre Pouget's research group at the University of Geneva (Switzerland) recently announced a solution to turn AI into a professional wine "sniffer" to distinguish the origin of wines, which can be applied to Limit scams.

Frauds about the origin of wine have long been a source of pain in the wine industry, causing losses to Europe of about 3 billion euros each year.

Professor Alexandre Pouget's research team used chromatography to analyze 80 wines from 7 different regions in Bordeaux, made over 12 years.

The AI ​​was taught by scientists to use algorithms based on all the chemicals detected in the wine to find identifiers instead of separating and identifying the compounds that make up the wine mixture.

AI will return results in a grid format. It will group wines with similar characteristics together. From there, based on the difference in concentration of the compounds, AI classifies wine according to each production region corresponding to the 7 survey locations.

AI returns results in almost real time, saving a lot of time compared to analyzing wine samples in the laboratory.

The research team said that developers can apply this technology to create applications that trace the origin of wine for customers to avoid fraud.

Professor Alexandre Pouget expects that the team will continue to research to make the system more complete, possibly adding other analytical functions through algorithms. For example, AI can determine the degree of blending of ingredients in a wine.

Isabella Humphrey

Update 05 December 2023