“Alexa, how do I smell?” could be one thing you could ask your virtual assistant in the future, thanks to two new studies.
In the first study, researchers from the University of California, Riverside developed an AI model capable of understanding what a chemical smells like. The model can also distinguish different smells based on the chemical makeup of the odor in question.
“We now can use artificial intelligence to predict how any chemical is going to smell to humans,” explained Anandasankar Ray, a cell and systems biologist at UC Riverside and co-author of the report. “Chemicals that are toxic or harsh in, say, flavors, cosmetics, or household products can be replaced with natural, softer, and safer chemicals.”
A new way to smell the roses
In the UC Riverside study, the team tried to model olfaction using chemicals and machine learning. Humans do it with the help of around 400 odorant receptors, which are triggered by a unique set of chemicals. When the receptors work together, a person is able to identify various smells—from freshly ground coffee to even your own fart.
“The power of machine learning is that it is able to evaluate a large number of chemical features and learn what makes a chemical smell like, say, a lemon or a rose or something else,” Ray added. “The machine learning algorithm can eventually predict how a new chemical will smell even though we may initially not know if it smells like a lemon or a rose.”
Beyond the ability to give your virtual assistant the ability to smell, the researchers noted that the algorithm can have practical applications in the food, flavor and fragrance industries. For one, the algorithm can help replace rare ingredients in perfume with existing and inexpensive chemicals that have a similar odor.
“It gives us a vast palette of compounds that we can mix and match for any olfactory application. For example, you can now make a mosquito repellent that works on mosquitoes but is pleasant smelling to humans,” said Ray.
The team developed an algorithm to learn which chemicals activate known human odorant receptors, using it to screen over half a million compounds for 34 odorant receptors. They then looked at whether the algorithm can also identify perceive the qualities of these chemical odorants.
The team found that not only was the algorithm able to predict the qualities of chemical odorants, it did so using fewer receptors.
“We now have an unprecedented ability to find ligands and new flavors and fragrances,” Ray added. “Using our computational approach, we can intelligently design volatile chemicals that smell desirable for use and also predict ligands for the 34 human ORs.”
Mapping how we smell
Olfaction, the scientific term for our sense of smell, has been poorly understood. While science can tell us more about how the brain process vision and hearing, it’s a little vague when it comes to smell. While scientists know that the nose communicates information about odor molecules to the brain’s olfactory bulb, how the brain sorts out these scents has been something that has eluded them.
In a separate study, neurobiologists from Harvard Medical School looked into the inner workings of the olfactory process and described how the brain encodes these different odors using machine learning.
The Harvard team showed how the brain categorizes smells using chemical similarities in the odors. In addition, they also noted that these categories are affected, even rewired, by other sensory experiences.
“All of us share a common frame of reference with smells,” explained senior author Sandeep Robert Datta. “You and I both think lemon and lime smell similar and agree that they smell different from pizza, but until now, we didn’t know how the brain organizes that kind of information.”