U.S. researchers just made another breakthrough in neuromorphic computing with their new synthetic synapse.
Physicists and engineers from the U.S. National Institute of Standards and Technology in Colorado have just developed a synthetic synapse that is deemed to be faster and more efficient than the real ones found in our brain. The team, headed by Mike Schneider, used superconducting technology to create their artificial neural connection.
In a paper published in the journal Science Advances, the researchers claimed that their superconducting switch might be the key to creating brain-like computers in the future. The synapse, which could be used to connect processors and store memories, can supposedly learn like its biological counterparts.@usnistgov researchers developed a synthetic synapse more powerful than the ones in our brain!Click To Tweet
“The synapse is widely believed to be integral for both learning and memory. With approximately 100 trillion synapses in the human brain, it is a critical component of neural circuitry. Hence, finding a simple, low-energy, artificial synapse is an important step in making a neuromorphic computer that can approach the level of complexity of the human brain,” the researchers wrote in their paper’s introduction.
What is a Synapse?
A Synapse, which can also be called a neuronal junction, is the site of transmission of electric nerve impulses between two nerve cells (neurons). There are also synapses between neurons and gland or muscle cells.
These electrical synapses enable neurons to communicate with each other. Ions, in this case called neurotransmitters, flow between cells through gap junctions. When these ions reached their target neurons, they fit into tailor-made receptors found on the surface of these neurons. This process then converts the chemical signals back into electric nerve impulses.
Have you ever wondered how fast the brain processes the signals that travel across over 100 trillion synapses situated between 100 billion neurons?
On average, a neuron could fire signals 50 times per second. So, if we do the math, 100 billion neurons multiplied by 50 firings would be equivalent to 50,000,000,000,000 (50 billion) signal firings per second.
Again, that’s 50 billion bits of information being processed simultaneously by the brain per second! Now, just imagine if you can harness this computing power into a computer. It would forever change the landscape of artificial intelligence and neuromorphic computing.
Creating a Synthetic Synapse
Schneider and his team were able to develop their synthetic synapse using superconducting materials that could transmit, process, and store information in magnetic flux units. A far cry from the electric current being utilized in conventional computers today.
The new neuromorphic hardware is said to be housed within a metallic cylinder which is around ten micrometers in diameter. It also has a layer of superconducting material with an insulator in between called a Josephson junction.
“These new artificial synapses are compatible with single flux quantum (SFQ) Josephson junction (JJ) circuits that can provide the underlying technology platform needed to scale large neuromorphic systems.”
Through its insulator fillings, the JJ circuits calibrate voltage standards by breaking down the current that passes through them to produce voltage spikes. This process is said to be based on a flexible internal design that can be customized through experience or by its environment. This also leads to stronger connections when more firing is made between two processors.
The synthetic synapse created by NIST is reportedly even better than the synapses found in our brain. Apparently, the NIST synapse could fire much faster, at around 1 billion times per second. Imagine, 1 billion firings per second against our brain’s 50 firings per second. That’s about one ten-thousandth as much as a NIST synapse could handle.
Furthermore, multiple NIST synapses could be 3D stacked to make large computing systems. With its small size, super fast-spiking signals, low energy requirements, and 3D stacking functionalities, the team’s synthetic synapse could pave the way for more complex neuromorphic computers to be created in the future.