What if computers worked like the human brain?
The era of the Internet of Things – the network of devices that allows them to share data – is already here. How should we handle the explosion of data and the increasing demand for energy? Computing hardware and software combinations require tens of kilowatts to compute, store, and make decisions using artificial intelligence, compared to 12 watts for the human brain. Inspired by this energy-efficiency, we look to mimic the way the brain’s neural networks process information through electrical impulses. Such a solution could operate trillions of sensors and cater to a growing multi-billion euro industry. Our advancement is ‘memristors’, carbon-based ferroelectric tunnel junction memory chips which work at room temperature, consume very little power, operate switches in nanoseconds, retain data for years even after being switched off, and fit the physical dimensions of current circuit boards.
Sayani Majumbar, Academy Research Fellow