Scientists Develop Powerful Microchip Modeled After The Human Brain
A microchip simulated after the brain may one day be able to move prosthetic limbs with the adeptness of the human brain.
Stanford scientists used the model of the human brain to develop an energy-efficient microchip that is 9,000 times faster than the average household computer. The development has opened a door to vast possibilities in robotics advancements and a greater understanding of the human brain.
The typical personal computer requires 40,000 times power to run than an average mouse's brain. Associate professor of bioengineering at Stanford, Kwabena Boahen and his team developed Neurogrid with power efficiency in mind in order to simulate the synaptic connections in the human brain.
Like Us on Facebook
The Neurogrid's circuit board is composed of 16 "Neurocore" chips, each of which support 65,536 neurons. For a long time scientists believed that there were over 100 billions neurons in the human brain, however in 2009, scientists counted and found there were an average of 86 billion. The grids were developed on a grant from the National Institutes of Health in order to improve upon the knowledge and research development efforts to better understand the brain, which is an ongoing effort of discovery.
The long-term goal for the team is to create an affordable yet power-efficient circuit board. Currently the Neurogrid costs about $40,000 and Boahen believes future cost reductions are a feasible reality. If an affordable, effective and power-efficient computer chip that was comparable to the human brain could be developed, it could be used to controlprosthetic limbs with the same speed and adeptness as average human actions.
"From a pure energy perspective, the brain is hard to match," says Boahen who wrote a paper on Neurogrid's capabilities.
The strategy they have approached the Neurogrid's development with is to enable the 16 Neurocore chips with their own unique synapses and have them share the same hardware circuits. The synapse is what triggers a neuron to release a neurotransmitter, which communicates to the next neuron. It is how a thought processes through the brain and produces a command in the body such as lifting an arm. As a whole, the Nuerogrid device is about the size of an iPad that can simulate brain commands on the same amount of power it takes to run an average tablet.
"The human brain, with 80,000 times more neurons than Neurogrid, consumes only three times as much power," Boahen writes. "Achieving this level of energy efficiency while offering greater configurability and scale is the ultimate challenge neuromorphic engineers face."
If scientists could use modern manufacturing processes that are used to fabricate the chips in large volumes, Boahen says he could cut a Neurocore chip's cost 100-fold. This would lead them down the road to eventually create a $400 copy of a million-neurons per circuit board. The more they can figure out how to decrease the costs, the more possibilities of using these powerfully-compact little microchips for other applications in medicine and technology.
Other efforts to create microchips modeled after the human brain have been made. The Systems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) Project is an attempt to recreate the brains ability to make synaptic connections. The ability to simulate the brain's ability to make connections would make a chip powerful enough to solve problems as proficiently as a human.
Heidelberg University's BrainScales project wants to, not only recreate synaptic behaviors, but neuron behaviors as well. Their High Input Count Analog Neural Network (HICANN) chip would allow researchers to simulate drug interactions in a shorter period of time without needing the use of a human brain.
These research and development goals are being improved upon quickly, however Neurogrid is currently the most cost-effective way to simulate neurons in comparison. Eventually, Boahen hopes, his team will develop a system affordable enough to be widely used in research and even to those that don't have a basis knowledge of bioengineering.
"Right now, you have to know how the brain works to program one of these," said Boahen, gesturing at the $40,000 prototype board on the desk of his Stanford office. "We want to create a neurocompiler so that you would not need to know anything about synapses and neurons to able to use one of these."
© 2012 iScience Times All rights reserved. Do not reproduce without permission.