Kiwi scientists set to unveil next generation of computing power.

Kiwi scientists will this week launch next-generation technology modelled on the greatest supercomputer ever designed - the human brain.

At the centre of a global conference, hosted at Auckland University of Technology (AUT), will be the NeuCube - an innovation that could allow us to predict natural disasters in real time, warn us of coming strokes, improve treatment for such conditions as schizophrenia, or control robots through brain signals.

The potential for these dramatic advances and many more lie in what's called neuro-computing - an emerging field in which scientists are moving away from the conventional computers of today, and toward machines that would process information like our own brains.

The largest existing device, based in the UK and known as the SpiNNaker (Spiking Neural Network architecture), uses specially-designed ARM processors to power software that could mimic the behaviour of potentially millions to billions of human neurons.


The problem is, said Professor Nik Kasabov, nobody yet knows how to unlock the full capabilities of such technology for efficient problem solving.

"We have developed a novel computational framework that can create flexible and configurable application systems to run on this super-computer hardware - or on any other neuromorphic chips or systems," said Professor Kasabov, director of AUT's Knowledge, Engineering and Discovery Research Institute (KENRI).

"And the developed application system, using our framework, can run on a simple laptop with an electronic board connected to it, or on a small chip embedded in a robot."

The NeuCube neurocomputer he originated draws upon the same information-processing principles in our brains, where information is represented in temporal sequences of electrical signals, or "spikes".

The machine - which could become the cornerstone of technology replacing laptops - combined software and hardware to tackle complex problems never solved before.
For humans, the potential benefits were as vast as they were startling.

"If we can use this neurocomputer to capture the patterns that are building up inside the human brain before a stroke happens, or before a patient of a neurodegenerative disease is treated with a drug, we can predict and prevent fatal events."

Dr Kasabov's team had already compared brain scans of people who had a stroke with those of healthy brains, finding the NeuCube was shown to be able to predict with 95 per cent accuracy whether someone was going to have a stroke in the next 24 hours.

As a feasibility analysis, a NeuCube application model was trained on retrospective New Zealand seismic data to predict risk of earthquakes early and accurately.


Its preliminary testing results were encouraging - the model was able to predict most of the large earthquakes around Christchurch 24 hours before the events.

The NeuCube neurocomputer will be launched before more than 80 leading scientists at the 13th Neuro-Computing and Evolving Intelligence, held this Thursday and Friday.

"We are very proud that we are leading the world with the NeuCube as the first neurocomputer of its kind."

But moving the NeuCube from patent-protected prototype to product would require hundreds of thousands of dollars of investment - and if New Zealand wanted to reap the benefits of the IP, the funds would need to come from here.

•People can check out the NeuCube for themselves in a demonstration to be held during the at AUT's WG Sir Paul Reeves Building Building, Governor Fitzroy Place, between 4pm-6pm on Thursday and Friday.

Brain power


• The brain contains an estimated 80 billion information-processing cells, each acting as a tiny electrical microprocessor that fires signals hundreds of times a second through a vast network of connections.

• Many scientists believe harnessing this processing power will allow us to meet the demands of a future explosion in new data and information.

• Scientists from AUT's Knowledge, Engineering and Discovery Research Institute believe they have found a new computational paradigm.

• The prototype device has retrospectively predicted earthquake and stroke risks.