Ale-gorithm a bottler of an idea for brewers

Thousands of years after the Sumerians first tinkered with “the divine drink”, brewers have turned to artificial intelligence (AI) for quality control of their flagship beers.

Australian Catholic University computational intelligence expert Niusha Shafiabady has developed a machine learning program that can predict the quality of beer without sampling a drop.

Designed in collaboration with a major brewer, the neural network considers 36 production parameters, including alcohol content and bitterness, before predicting the brew’s caliber.

“It takes out the guess work,” Associate Professor Shafiabady said of the fresh take on the artisanal brewing process thought to have originated with the Sumerians more than 4000 years ago.

Associate Professor Shafiabady built nine different models to predict a beer’s pH and Nibem (stability of the head, or foam) levels before water is even introduced to the malt.

Each model analyses the impact on pH and Nibem of the cocktail of parameters that also includes, CO2, colour and temperature, haziness, pressure, water volume and what varieties of hops and malt are used.

The machine learning programs could be trained to reliably predict the beer’s acidity and head quality.

“Instead of tweaking those variables manually, they can find out what those values should be ahead of production to ensure the desired quality,” she said.

“That changes the game for beer producers because they will know beforehand the optimal combination of those variables to produce the best outcome.”

Media Contact: Damien Stannard, 0484 387349, damien.stannard@acu.edu.au

Have a question?

We're available 9am–5pm AEDT,
Monday to Friday

If you’ve got a question, our AskACU team has you covered. You can search FAQs, text us, email, live chat, call – whatever works for you.

Live chat with us now

Chat to our team for real-time
answers to your questions.

Launch live chat

Visit our FAQs page

Find answers to some commonly
asked questions.

See our FAQs