Discover How Algorithms Identify American vs. Scotch Whiskey Origins

Algorithms Can Distinguish Between American and Scotch Whiskey(Source: Phys.org)

Introduction

Recent research has demonstrated the capability of two machine learning algorithms to accurately classify whiskey based on its origin—American or Scotch—and to identify its strongest aroma notes.

Research Findings

• The algorithms were able to classify whiskeys with over 90% accuracy.
• Key aroma compounds were linked to each classification:
  • American Whiskey: Menthol, citronellol (caramel-like aroma)
  • Scotch Whiskey: Methyl decanoate, heptanoic acid (apple-like, solvent-like, and phenolic aromas)
• Both algorithms surpassed human experts in identifying the five strongest aroma notes.

Methodology

The study involved analyzing the molecular composition of seven American and nine Scotch whiskies using:

• Gas chromatography
• Mass spectrometry

The data generated was then processed through the algorithms developed by the researchers, including a specialized molecular odor prediction algorithm named OWSum and a neural network.

Significance of the Study

This research highlights the potential of using artificial intelligence for quick classification and identification of whiskey aromas, which could significantly reduce the time and resources required compared to traditional sensory analysis involving human panels.

Conclusion

The findings affirm that algorithms can effectively assess complex mixtures of odorous compounds in whiskey, paving the way for advancements in the food and beverage industry through enhanced classification technologies.