A computer that works like the human brain? It may not be that far off.
IBM cognitive computing researchers are one step closer to making a brain-like computer a reality. The team on Thursday unveiled a "breakthrough software ecosystem" that has the ability to program microchips with architecture based on the human brain.
The framework, which includes new programming language and algorithms, is just part of the process that principal investigator Dr. Dharmendra S. Modha hopes will lead to IBM's long-term goal:
"Our end goal here is literally to build a brain in a box with 100 trillion synapses," Modha told The Huffington Post. (That's about the same number of synapses within the human brain.)
Led by Modha, the team has worked on the SyNAPSE project since 2008, mapping out the human brain in order to reproduce its structure and architecture. Modha described the multi-institutional effort as an amalgamation of neuroscience, computing and nanotechnology.
Though researchers are seeking to create a brain-like computer capable of bringing computation and intelligence to data, instead of vice versa, their goal is not to do away with standard computing systems altogether.
"We're not trying to replace today's computers. We're trying to compliment them," Modha said, comparing standard systems to the analytical left side of the brain, and the team's synaptic model to the more intuitive right brain.
Typically, computers accept data and transport it for processing. However, the SyNAPSE architecture functions like a network of neurosynaptic cores, all of which are interconnected yet able to function independently.
"The sensor becomes the computer, essentially," Modha explained to HuffPost.
First introduced in 2011, each neurosynaptic chip consists of 256 "neurons," used for computation, and 256 "axons" for communication, similar to the human brain. In addition, the "neurons" and "axons" are connected by 65,536 "synapses," so each neuron is connected to every axon. Thus, the computer chip is able to mimic the brain's abilities of perception, action and cognition.
Ultimately, the team hopes to expand the network to include 10 billion neurons and 100 trillion synapses, while also scaling down power consumption as well as size so that the system could fit in a shoebox. Having recently received $12 million in federal funding from the U.S. Department of Defense for the fourth phase of the project, the team will now refine the architecture and attempt to improve the software and hardware.
"It's really a long marathon," Modha said, "and we're running fast."
Watch the video below to learn more about the new developments in the SyNAPSE project.
CORRECTION: An earlier version of this article erroneously stated that the neurosynaptic chip consists of 265 neurons. The correct number is 256.