Humans have the ability to recognize over a trillion odors. If AI technology can recognize even 1% of these odors, it will be an extraordinary advance in the field of machine learning, since it is very difficult to teach it to distinguish smells.
Now, however, researchers from Intel and Cornell University have managed to take a step in that direction. They created a new neuromorphic chip, called Loihi, that has learned to recognize the smell of at least 10 different dangerous chemicals.
As published in the journal Nature Machine Intelligence, the researchers claimed that the chip can detect specific odors.
The accuracy demonstrated by this chip is astonishing, since, as the researchers report, machine learning would normally need at least 3,000 more samples to achieve this level of accuracy.
Intel Loihi Chip
Intel's neuromorphic chip includes over 2 billion transistors and 130,000 artificial neurons and, as Intel, is 1,000 times faster and 10,000 times more efficient at processing information than conventional processors.
Neuromorphic chips, like Loihi, are designed to be able to replicate the properties of the complex biological architecture of the nervous system.
Intel's Loihi chip has managed to mimic the part of the brain responsible for recognizing odors and the sense of smell.
Nabil Imam, a senior researcher at Intel, said that the creation of this chip is a significant leap in technology, since it can help in the future in detecting weapons and explosives, discovering drugs, and even diagnosing diseases.
This particular project is an example of what the combination of neuroscience and artificial intelligence can achieve and demonstrates the ability of a chip to successfully emulate important capabilities that can benefit many industries in the future.
In addition to Intel, other companies are trying to train their AI systems to recognize odors. Researchers at Google Brain are trying to train their AI technology to recognize the smell of objects, but based on their molecular structure.
