Researchers at the University of California, Santa Barbara have developed a functional artificial neural circuit of about 100 artificial synapses, or junctions between two nerve cells, proven to perform a simple version of image classification.
The human brain remains a model of computation and efficiency for engineers like Dmitri Strukov and his colleagues in University of California, Santa Barbara’s Department of Electrical and Computer Engineering, Mirko Prezioso, Farnood Merrikh-Bayat, Brian Hoskins, and Gina Adams. The brain can accomplish certain functions in a second that computers would require far more time and energy to perform. Some of these features include making split-second decisions about the letters and symbols you see, classifying shapes and positions relative to each other, and deriving different levels of meaning through many channels of context.
In the researcher’s demonstration, the circuit implementing the artificial neural network was able to classify three letters, z, v, and n, by their image. In a process similar to how humans pick out friends from a crowd or find the right key from a ring of keys, the simple neural circuitry was able to correctly classify the simple images.
“It’s a small, but important step,” said Dmitri Strukov, a professor of electrical and computer engineering. With time and further progress, the circuitry may eventually be expanded and scaled to approach something like the human brain’s, which has 10^15 (one quadrillion) synaptic connections.
The key to this technology is the memristor, an electronic component whose resistance changes depending on the direction of the flow of the electrical charge. The memristor operation is based on ionic movement, similar to the way human neural cells generate electrical signals.
“While the circuit was very small compared to practical networks, it is big enough to prove the concept of practicality,” said Merrikh-Bayat. “And, as more solutions to the technological challenges are proposed, the technology will be able to make it to the market sooner.”
“The memory state is stored as a specific concentration profile of defects that can be moved back and forth within the memristor,” said Strukov. The ionic memory mechanism brings several advantages over purely electron-based memories, which makes it very attractive for artificial neural network implementation, he added.
“For example, many different configurations of ionic profiles result in a continuum of memory states and hence analog memory functionality,” he said. “Ions are also much heavier than electrons and do not tunnel easily, which permits aggressive scaling of memristors without sacrificing analog properties.”
In order to create the same human brain-type functionality with conventional technology, the device would have to be loaded with multitudes of transistors that would require far more energy. To be able to approach functionality of the human brain, many more memristors would be required to build more complex neural networks to do things we do with little effort, such as identify different versions of the same thing or infer identify of an object.
“Classical computers will always find an ineluctable limit to efficient brain-like computation in their very architecture,” said lead researcher Prezioso. “This memristor-based technology relies on a completely different way inspired by biological brain to carry on computation.”
Potential applications already exist for this technology, such as medical imaging, improvement of navigation systems, or searches based on images rather than on text. The compact circuitry the researchers are creating would also go a long way toward creating the kind of high-performance computers and memory storage devices users will continue to seek long after the proliferation of transistors.
“The exciting thing is that, unlike more exotic solutions, it is not difficult to imagine this technology integrated into common processing units and giving a serious boost to future computers,” said Prezioso. “There are so many potential applications—it definitely gives us a whole new way of thinking,” he said.
The researchers are currently improving the performance of the memristors by scaling the complexity of circuits and enriching the functionality of the artificial neural network. The next step is to integrate it with semiconductor technology in order to enable more complex demonstrations and allow the early artificial brain to do perform more complicated tasks.