The Holy Grail of computing? While it is true that the human brain is used in a comparative way to know how a machine works, the truth is that the operation of computers and human brains is not at all alike. This could change thanks to computer technology that uses light to mimic the functionality of neurons’ synapses, paving the way for hardware that combines the speed of modern processors with the efficiency of a computer’s brainpower. Ready for the brain of the future?
Computers combine electrical impulses with small on and off switches for their functions; In neurons, on the other hand, chemistry is what distributes impulses through the channels that we know as synapses (connections between neurons). The difference is considerable in terms of memory and power consumption, as no hardware comes close to the efficiency and storage capacities of a human brain.
Now, a team of researchers from the Universities of Oxford, Exeter (United Kingdom) and Münster (Germany), has managed to go one step further by creating a photonic integrated circuit that acts like a synapse.
“The development of computers that function like the human brain has been a holy grail of scientists for decades. Through a network of neurons and synapses, the brain can process and store enormous amounts of information simultaneously, using a few tens of watts of power. Conventional computers cannot approach this type of performance, “explains Harish Bhaskaran, co-author of the work published in the journal Science Advances.
In computers, there are processor units that control logic and memory. E n the human brain, do not have a CPU or a hard drive, it is chemical processes.
And, depending on factors such as wave strength or frequency, neurotransmitter surges can continue the message by jumping to other neurons, stopping or speeding up the signal.
Described as synaptic plasticity, changes in this checkpoint can explain how we learn and process new information, strengthening some circuits while allowing others to wither away.
The so-called neuromorphic engineering aims to replicate this way of combining processing and memory in a system, bringing biology and artificial intelligence even closer, that is, that microprocessors can be configured more like the human brain than traditional computer chips.
The key has been to build a processor that does the same thing as synapses. “Since synapses outnumber neurons in the brain by about 10,000 to 1, any brain machine must be able to replicate some form of synaptic mimicry. That’s what we’ve done here, ” says Wolfram Pernice, co-author of the investigation.
Artificial synapse or synaptic emulation is based on structures made of a phase change material ( PCM), which stores and releases significant amounts of energy as it changes from one state to another (solid ↔ liquid).
Light waves are channeled through the material, with optical pulses changing the PCM in such a way that it mimics the plasticity of a synapse. Although the concept is not new, it is the first time that it has been put into practice.
Light-based neuromorphic processors seem to be the perfect match between mind and machine. We attend the future of artificial intelligence and computing.
Reference: On-chip photonic synapse Zengguang Cheng, Carlos Ríos, Wolfram HP Pernice, C, David Wright, Harish Bhaskaran. Science Advances 2017 DOI: 10.1126 / sciadv.1700160
