Machine learning and artificial intelligence are finding applications in many different areas but scientists from Belgium have now raised the, er, bar somewhat.
They have used machine-learning algorithms to predict the taste and quality of beer and what compounds brewers could use to improve the flavour of certain tipples.
Kevin Verstrepen from KU Leuven and colleagues spent five years characterizing over 200 chemical properties from 250 Belgian commercial beers across 22 beer styles, such as Blond and Tripel beers. They also gathered tasting notes from a panel of 15 people and from the RateBeer online beer review database.
They then trained a machine-learning model on the data, finding it could predict the flavour and score of the beers using just the beers’ chemical profile.
By adding certain aromas predicted by the model, the researchers were even able to boost the quality – as determined by blind tasting – of existing commercial Belgian ale.
The team hops the findings could be used to improve alcohol-free beer. Yet KU Leuven researcher Michiel Scheurs admits that they did celebrate the work “with the alcohol-containing variants”.
