Scientists have created a circuit board modelled on the human brain that is 9,000 times faster and dramatically more power efficient than average PC.
By combining advanced computational capabilities with low power consumption, scientists at Stanford University have created a circuit board that could eventually drive next-generation prosthetic limbs, offering significant advancements on existing chip technology.
The board, known as Neurogrid, is an example of the growing field of neuromorphic computing; the practice of modelling electronic circuits on the structure of the brain and nervous system, the most advanced computer present in nature.
Neurogrid consists of 16 custom Neurocore chips, which when combined can simulate a million neurons and several billion synaptic connections – the connectors between brain cells that allow signals to be sent between them.
While this is an impressive achievement, it represents only a fraction of the computational power of the human brain, which has an average of 86 billion neurons.
Nevertheless, Neurogrid represents the pinnacle of neuromorphic research, particularly when power consumption is taken into account. The circuit board is around the size of an iPad and requires a similar amount of power, putting it leaps ahead of its competitors when it comes to power output vs energy consumption.
This power efficiency is extremely important in clinical uses as it allows Neurogrid to be incorporated into medical prosthetics and other life-enhancing wearable technologies.
Potentially the technology could be used to operate prosthetics for paralysed users, and if developed it could eventually be used to control such prosthetics as quickly and carefully as we control our own limbs.
However, the technology still faces a major roadblock that must be overcome if it is to be used in this way. At present, programming Neurogrid requires a detailed knowledge of neuroscience, which is an issue for scientists developing Neurogrid-controlled prosthetics.
Lead researcher and Stanford University associate professor of bioengineering Dr Kwabena Boahen is in the process of tackling this issue by creating a neurocompiler that would take over the task.
“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,” explained Boahen.
The field of neuromorphic research has a number of projects that could eventually lead to some remarkable breakthroughs for computing.
Among these is the Human Brain Project from the European Union, which is looking to simulate an entire human brain on a supercomputer.
Although in the early stages, this project has the potential to radically impact medicine, neuroscience and computing, potentially one day paving the way for computers that are as complex, or perhaps more so, than the human brain.
Images courtesy of Stanford University.