Companies are making strides in artificial intelligence (AI) and machine learning, and based on these technologies, Intel may just have made the coolest chip ever. Called Loihi, the chip can smell 10 hazardous chemicals. The company isn’t stopping there, and has scaled the chip up to Pohoiki Springs -- a neuromorphic research system that provides a computational capacity of 100 million neurons.
Researchers from Intel Labs and Cornell University jointly published a paper in Nature Machine Intelligence detailing the abilities of Intel’s neuromorphic research chip, Loihi, to learn and recognize hazardous chemicals in the presence of significant noise and occlusion. According to the paper, Loihi learned each odour with just a single sample, without disrupting its memory of previously learned scents. It took 3,000 training samples per class to reach a superior recognition accuracy.
‘We are developing neural algorithms on Loihi that mimic what happens in your brain when you smell something. This work is a prime example of contemporary research at the crossroads of neuroscience and artificial intelligence and demonstrates Loihi’s potential to provide important sensing capabilities that could benefit various industries,’ said Nabil Imam, Senior Research Scientist in Intel’s Neuromorphic Computing Lab.
To make the chip recognise a smell, the researchers used a dataset consisting of the activity of 72 chemical sensors in response to these smells and configured the circuit diagram of biological olfaction on Loihi. The chip quickly learned the neural representation of each of the smells and recognised each odour.
This is called Neuromorphic computing, in which scientists try to apply insights from neuroscience to create chips that function like the human brain. Neuromorphic systems replicate the way neurons are organised, communicate, and learn at the hardware level. One such example is Pohoiki Springs, the company’s most powerful neuromorphic research system.
‘Pohoiki Springs scales up our Loihi neuromorphic research chip by more than 750 times, while operating at a power level of under 500 watts. The system enables our research partners to explore ways to accelerate workloads that run slowly today on conventional architectures, including high-performance computing (HPC) systems,’ added Mike Davies, Director of Intel’s Neuromorphic Computing Lab.
Intel says that this cloud-based system will be made available to members of the Intel Neuromorphic Research Community (INRC), extending their neuromorphic work to solve larger, more complex problems.