By utilizing AI models, food and drink manufacturers can save a significant amount of time and money when developing new products or improving existing recipes to cater to consumer preferences. This can be achieved by training machine-learning models on a comprehensive data set that includes both chemical composition and sensory features of different foods and drinks.
Over the course of five years, researchers analyzed 250 commercial beers to measure their chemical properties and flavor compounds, which are key determinants of taste. The detailed chemical analyses were combined with assessments from a trained tasting panel and reviews from an online platform to create a comprehensive data set. This data set was then used to train 10 machine-learning models, which were designed to accurately predict a beer’s taste, aroma, mouthfeel, and overall quality, as well as predict how likely consumers would be to rate it highly.
By incorporating a wide range of data sources, the researchers were able to develop sophisticated models that could help enhance the beer production process. These models can be used by manufacturers to improve their products or develop new ones that meet consumer preferences more accurately. This can result in significant cost savings and improved efficiency in the production process. Overall, utilizing AI models in food and drink manufacturing has great potential for improving product development and enhancing consumer experience.
