“Predict and improve complex beer flavor through machine learning.” This is the title of a scientific article published in Nature Communications about the potential benefits of using Artificial Intelligence (AI) in creating better beers – or at least that are more aligned with consumer tastes.
For five years, a group of researchers analyzed 250 beers sold to the general public, measuring their chemical properties and flavor compounds. This data was then combined with the analysis carried out by professional beer tasters, to which data from more than 180 thousand analyzes made by consumers on the RateBeer platform were added, on elements such as taste, appearance, aroma and overall quality.
This data set was then used to train ten machine learning models with the aim of predicting the taste, smell and texture in the mouth, as well as the likely rating that a consumer would give a given beer. One of these models stood out – Gradient Boosting was able to predict with greater accuracy the rating an end user would give to a beer than human experts.
Furthermore, it was possible to identify specific compounds that increase users’ taste for beer. One of these cases is lactic acid, which helps give beers a ‘fresher’ flavor, explains the publication MIT Technology Review.
The researchers at the Leuven Microbiology Center, responsible for the study, actually carried out this experiment – they took identified components and added them to beers. “And to our amazement, as soon as we did the blind tests, the beers became better, they were generally more appreciated”, highlights researcher Kevin Verstrepen.
In addition to helping manufacturers to be able to produce more ‘optimized’ beers depending on a specific target audience, and in addition to having the potential to reduce the time it takes to create new beers, by reducing the time required in fine-tuning the product and in the of tastings, one of the major areas in which the use of Artificial Intelligence could have a major impact is in the world of beers… without alcohol. Typically, the different taste of beers with alcohol is one of the reasons that keeps consumers away from this category of drinks, but it will now be possible to add specific ingredients that make beers tastier for users.
“Our study reveals how big data and machine learning uncover complex links between food chemistry, taste and consumer perception, and lays the foundation for developing new, personalized foods with superior flavors,” emphasize the researchers in published study.
Tags: Exam mission Artificial Intelligence beer
