(eommina)
Predicting color perception is easy: specific wavelengths produce specific colors that most people see in a consistent way. But predicting how a particular molecule will smell is much tougher. So a group of researchers set up a contest and invited teams of computer scientists to come up with a set of algorithms able to predict the odor of different molecules based on their chemical structure. A team led by Yuanfang Guan, a computer scientist at U-M, scored best at predicting how individual subjects rate smells. Twenty-two teams from around the globe took part in the challenge, using a massive database of more than 1 million data points collected from a previous study. It remains to be seen how broadly useful an AI-powered nose will be, but it could help fragrance and flavor companies come up with new molecules customized to trigger particular smells.
