AI Identifies Whiskey Aromas and Origins with High Accuracy
Researchers demonstrate machine learning algorithms that predict whiskey flavor notes and country of origin more consistently than human experts.
Overview
- A study used two machine learning algorithms to analyze the molecular composition of 16 whiskies, distinguishing between American and Scotch varieties with over 90% accuracy.
- The algorithms identified key flavor notes in each whiskey more consistently than a panel of 11 human experts, matching aggregate results from the panel's assessments.
- American whiskies were associated with flavors like caramel and compounds such as menthol, while Scotch whiskies were linked to smoky or medicinal notes and compounds like methyl decanoate.
- The AI approach could enhance efficiency and consistency in whiskey quality control, ensuring flavor stability across batches and blends.
- Researchers suggest potential applications beyond whiskey, including detecting counterfeit products or improving odor management in recycled materials.