IBM Researchers Succeed In Creating Artificial Stochastic Phase-Change Neuron

IBM Researchers Succeed In Creating Artificial Stochastic Phase-Change Neuron

IBM researchers in Switzerland have created an artificial neuron that behaves just like the real thing. For the first time in history, artificial phase-change neurons have been grouped together (in a population of 500 synthesized in a lab) to process a neurological signal in more or less the same way that biological neurons transmit messages. They can be made exceptionally small and are similar in power and energy usage to biological neurons, and can even produce results with random variations, also just like biological neurons.

Scientists have wanted to create artificial brains for decades and have theorized about replicating the way the brain works by creating a large population of neurons, but until now it has not been possible due to the amount of power that would be required to pull off a result comparable to a biological brain.

To this end, for over a decade IBM’s researchers have been studying phase-change materials – a type of substance with a high heat of fusion that is capable of storing and releasing large amounts of energy when it solidifies and melts at a certain temperature.

The researchers have previously managed to figure

out how to store data in the phase-change memory for the first time, and now they are proving that it is possible to create randomly spiking neurons with the phase-change materials in order to store and process data.

Like a biological neuron, IBM’s artificial neuron has inputs (dendrites), a neuronal membrane (lipid bilayer) around the spike generator (soma, nucleus), and an output (axon). There’s also a back-propagation link from the spike generator back to the inputs, to reinforce the strength of some input spikes.

The key difference is in the neuronal membrane. In a real neuron, this would be a lipid bilayer, which essentially acts as both a resistor and a capacitor: it resists conductance, but eventually, with enough electricity along the input dendrite, it builds up enough potential that its own spike of electricity is produced—which then flows along the axons to other neurons—and so on and on.

In IBM’s neuron, the membrane is replaced with a small square of germanium-antimony-tellurium (GeSbTe or GST). GST, which happens to be the main active ingredient in rewritable optical discs, is a phase-change material. This means it can happily exist in two different phases (in this case crystalline and amorphous), and easily switch between the two, usually by applying heat (by way of laser or electricity). A phase-change material has very different physical properties depending on which phase it’s in: in the case of GST, its amorphous phase is an electrical insulator, while the crystalline phase conducts.

With the artificial neurons, the square of GST begins life in its amorphous phase. Then, as spikes arrive from the inputs, the GST slowly begins to crystallise. Eventually, the GST crystallises enough that it becomes conductive—and voilà, electricity flows across the membrane and creates a spike. After an arbitrary refractory period (a resting period where something isn’t responsive to stimuli), the GST is reset back to its amorphous phase and the process begins again.

“Stochastic” refers to a system where there is an amount of randomness in the results. Biological neurons are stochastic due to a range of different noises (ionic conductance, thermal, background). IBM says that its artificial neurons exhibit similar stochastic behaviour because the amorphous state of each GST cell is always slightly different after each reset, which in turn causes the crystallisation process to be different. Thus, the engineers never quite know exactly when each artificial neuron will fire.

The artificial neurons cannot store any digital information, but when electrical pulses where applied to the neurons, a crystallization process began that resulted in the phase-change material eventually firing, which replicates a biological function in neuroscience called the “integrate and fire” property of biological neurons, such as the immediate response triggered by the brain when humans touch something hot.

Using this method to get the artificial neurons to fire, each neuron can then be used to detect patterns and notice correlations in streams of data being collected from IoT-connected devices, such as providing analytics about how smart appliances in the home are being used.

The researchers organised hundreds of the artificial neurons into populations and used them to represent fast and complex broadband signals like in the brain, building five arrays of neurons that each contained 100 neurons each.

It takes 60 million microwatts to power a 60-watt lightbulb, but in comparison, the energy required for each neuron update was less than five picojoule and the average power is less than 120 microwatts. The researchers also found that the artificial neurons could sustain billions of switching cycles, meaning that they will last a really long time, potentially multiple years, if the update frequency is 100Hz.

Phase-change artificial neurons already exist today but they are each 90 nanometres in size. IBM’s researchers told that they can potentially shrink the phase-change artificial neurons to just 14 nanometres, which means that neuromorphic computer processors might one day be possible.

“We show we can have both synapses and neurons using phase-change cells,” said Tomas Tuma, lead author of the paper and a scientist at IBM Research in Zurich. “The discovery is important in taking phase-change memory to the next level and to use it for computations.”

The research, entitled Stochastic phase-change neurons is published in the journal Nature Nanotechnology.