J. M. COHEN (1), Z. B. J. Bushman (1), Y. Zhou (1); (1) Analytical Flavor Systems, State College, PA, U.S.A.
—This poster was recipient of a 2015 Best Poster Award
A successful quality program in a brewery must be capable of detecting flaws, taints, contaminations, and batch-to-batch deviations. GC/MS is considered the gold standard of quality control in brewing but requires capital investment, human experts, and constant upkeep. Analytical Flavor Systems has developed a machine-learning and artificial intelligence-based system capable of detecting and predicting quality control problems using only 16 sensory attributes and 8 sensations from human sensory data, without the need for expensive instruments. The sensory data generated by individual panelists can be used to identify a flaw, taint, or contamination, in real time, even if the taster is unable to recognize the flaw themselves. The sensory data is used to build flaw-detecting machine-learning algorithms, by training the algorithm on human-generated sensory data representing several different styles of beer spiked with 20 different contaminants. The Gastrograph review mobile application was used to collect sensory data, and the data were then used to construct models. These models were then tested in a similar fashion. The models were shown to be able to detect and identify, with a high degree of accuracy, an array of flaws found in beers.
Jason M. Cohen is the founder, CEO, and lead data scientist of Analytical Flavor Systems (AFS). Before starting AFS, Jason was the founder and executive director of The Tea Institute at Penn State, which oversees more than 20 researchers in 5 fields of study in traditional Chinese, Japanese, and Korean teas. Jason did his research in sensory science and data mining, developing the Gastrograph system after more than 3 years of research. Jason is a professional coffee, tea, and beer taster, and when he’s not trying new products, he enjoys rock climbing, ice climbing, and fencing.