Can a computer detect and recognize specific odors? This is the direction in which a research project led by Intel is Cornell University with the common goal of demonstrating how the neuromorphic computing is paving the way for the so-called smell science.
Can a chip recognize odors?
The aim is to arrive one day to fine-tune electronic detection systems capable, for example, of sniffing the presence of explosives, narcotics, gases and pathogens. The study was published on Nature Machine Intelligence.
At the moment, during the experimental phase of the project, the neuromorphic chip Loihi Intel has been trained to identify hazardous chemicals by subjecting only one sample to its analysis and without damaging the memory containing those previously recognized. These are the words of Nabil Imam (in the photograph below), Senior Research Scientist of Intel Neuromorphic Computing Lab.
We are developing neural algorithms with Loihi that mimic what happens in the brain when we smell something. This work is the first example of contemporary research that intersects neuroscience and artificial intelligence and demonstrates Loihi's potential to offer important sensory skills from which various sectors will benefit.
The results that have emerged so far are encouraging:accuracy it is higher than that of other methods developed with the same purpose, also using deep learning.
To understand which perspectives could open projects of this type, which applications could give life to the intersection between artificial intelligence is neuroscience, we refer to the reading of the article "How a computer chip can smell without having a nose" published on Intel's official blog, while in the streaming video above an explanation of what neuromorphic computing is: the attempt to instruct a computer to think and process data in a way similar to that in which our brain operates.