September 24, 2016
Scientists at IBM have claimed a computational breakthrough after imitating large populations of neurons for the first time.
Neurons are electrically excitable cells that process and transmit information in our brains through electrical and chemical signals. These signals are passed over synapses, specialised connections with other cells.
It’s this set-up that inspired scientists at IBM to try and mirror the way the biological brain functions using phase-change materials for memory applications.
“The breakthrough marks a significant step forward in the development of energy-efficient, ultra-dense integrated neuromorphic technologies for applications in cognitive computing,” the scientists said.
The artificial neurons consist of phase-change materials, including germanium antimony telluride, which exhibit two stable states, an amorphous one (without a clearly defined structure) and a crystalline one (with structure). These materials are also the basis of re-writable Blue-ray but in this system the artificial neurons do not store digital information; they are analogue, just like the synapses and neurons in a biological brain.
The beauty of these powerful phase-change-based artificial neurons, which can perform various computational primitives such as data-correlation detection and unsupervised learning at high speeds, is that they use very little energy – just like human brain.
In a demonstration published in the journal Nature Nanotechnology, the team applied a series of electrical pulses to the artificial neurons, which resulted in the progressive crystallisation of the phase-change material, ultimately causing the neuron to fire.
In neuroscience, this function is known as the integrate-and-fire property of biological neurons. This is the foundation for event-based computation and, in principle, is quite similar to how a biological brain triggers a response when an animal touches something hot, for instance.
As part of the study, the researchers organised hundreds of artificial neurons into populations and used them to represent fast and complex signals. When tested, the artificial neurons were able to sustain billions of switching cycles, which would correspond to multiple years of operation at an update frequency of 100Hz.
The energy required for each neuron update was less than five picojoule and the average power less than 120 microwatts — for comparison, 60 million microwatts power a 60 watt light bulb, IBM’s research paper said.
When exploiting this integrate-and-fire property, even a single neuron can be used to detect patterns and discover correlations in real-time streams of event-based data. “This will significantly reduce the area and power consumption as it will be using tiny nanoscale devices that act as neurons,” IBM scientist and author, Dr. Abu Sebastian told WIRED.
This, IBM believes, could be helpful in the further development of internet of things technologies, especially when developing tiny sensors.
“Populations of stochastic phase-change neurons, combined with other nanoscale computational elements such as artificial synapses, could be a key enabler for the creation of a new generation of extremely dense neuromorphic computing systems,” said Tomas Tuma, co-author of the paper.
This could be useful in sensors collecting and analysing volumes of weather data, for instance, said Sebastian, collected at the edge, in remote locations, for faster and more accurate weather forecasts.
The artificial neurons could also detect patterns in financial transactions to find discrepancies or use data from social media to discover new cultural trends in real time. While large populations of these high-speed, low-energy nano-scale neurons could also be used in neuromorphic co-processors with co-located memory and processing units.