Photo credit: phys.org
Machine Learning Unveils Whiskey Origins and Aromas
Recent research published in Communications Chemistry demonstrates the capabilities of two machine learning algorithms in distinguishing between American and Scotch whiskies while also identifying their most prominent aromas. This study indicates that these algorithms can perform better than human experts when it comes to recognizing key aromatic notes in whiskey.
The olfactory profile of whiskey arises from a complex array of volatile compounds, making it difficult to assess its aromatic characteristics based solely on chemical composition. Traditional methods often rely on panels of trained experts, a process that is resource-intensive in terms of time and finances, and often leads to variable consensus among tasters.
In their study, researchers led by Andreas Grasskamp evaluated the molecular makeup of seven American and nine Scotch whiskies. They utilized two distinct algorithms: OWSum, a molecular odor prediction tool created by the research team, and a neural network. The molecular data was sourced from established gas chromatography and mass spectrometry analyses, which are standard techniques for separating and identifying mixtures.
The algorithms were tasked with determining the whiskey’s origin and pinpointing its five most dominant aromatic notes. The findings showed that OWSum could accurately classify whiskey as either American or Scotch over 90% of the time. Notably, certain compounds were strongly linked to each classification: menthol and citronellol with American whiskey, while methyl decanoate and heptanoic acid favored Scotch.
OWSum highlighted caramel-like aromas as the defining feature of American whiskies, whereas Scotch whiskies were characterized by apple-like, solvent-like, and phenolic notes, frequently described as smoky or medicinal. Remarkably, both algorithms surpassed individual human experts in accurately and consistently identifying the strongest aromatic notes of specific whiskies.
The implications of this research suggest a future where quick classification of whiskies and precise identification of their key aromatic notes could be achieved through algorithmic approaches, potentially transforming the field of whiskey tasting and evaluation.
More information: Andreas Grasskamp, Odor prediction of whiskies based on their molecular composition, Communications Chemistry (2024). DOI: 10.1038/s42004-024-01373-2. www.nature.com/articles/s42004-024-01373-2
Citation: Algorithms can determine whether a whiskey is of American or Scotch origin (2024, December 19) retrieved 19 December 2024 from https://phys.org/news/2024-12-algorithms-whiskey-american-scotch.html
Source
phys.org