Stanford Engineers Create Circuit Board that Mimics Brain [Video]

Engineers at Stanford University have created a circuit board fashioned after the human brain. The new circuit board is 9,000 times faster than current computers and uses far less power.

The research on "artificial brains" could pave way for next-generation computers and other electronic devices.

Modern computers, as sophisticated as they seem, are no match for the brain designed by nature. A simple brain region of a mouse is 9,000 times faster than an average PC. Also, computers use 40,000 times more power, according to a news release from the University.

"From a pure energy perspective, the brain is hard to match," said  Kwabena Boahen, associate professor of bioengineering at Stanford.

Boahen's latest article in Proceedings of the IEEE, describes how "neuromorphic" researchers in the U.S and Europe are building intelligent machines using designs perfected by nature.

The European Union's Human Brain Project wants to build an artificial brain using a supercomputer. U.S. BRAIN Project, too, wants to simulate a brain, but is bent towards developing tools that could read and write signals like the brain.

Boahen and team created a circuit board called Neurogrid, which consists of 16 Neurocore chips. These chips mimic the function of millions of neurons and synaptic connections in the brain. The Neurogrid is a device about the size of an iPad and copies the function of a brain.

Devices running with the speed and power efficiency of a human brain could help design new types of prosthetics. A Neurocore-like chip could use signals from the brain of an amputee and send commands to the artificial limb.

The research team now plans on lowering the cost of their Neurogrid. Each of these devices takes around $40,000 to manufacture. The team manufactured the device using 15-year-old fabrication methods.

According to Boahen, switching to advanced manufacturing process could bring down the price of the circuit board to $400.

Of course, the chip is slower than a human brain.

"The human brain, with 80,000 times more neurons than Neurogrid, consumes only three times as much power," Boahen wrote in his article. "Achieving this level of energy efficiency while offering greater configurability and scale is the ultimate challenge neuromorphic engineers face."