Brain-like machines offer hope

Renowned scientist Professor Nik Kasabov. Photo / NZME.

Creating computers that "think" as the human brain does could pave the way for dramatic breakthroughs, ranging from predicting strokes to predicting natural disasters in real time.

The possibilities of what's called neuro-computing - an emerging field in which scientists are moving away from today's conventional computers to machines that could process information like our own brains - are as many as they are fascinating.

This week, New Zealand boosted its efforts in the field with the opening of AUT University's state-of-the-art neurocomputing lab - the NeuLab.

Heading a programme of next-generation research there is renowned scientist Professor Nik Kasabov, whose innovation called NeuCube has the potential to process complex spatio-temporal data related to improving treatment for such conditions as schizophrenia, or controlling robots through brain signals.

It drew upon the same information-processing principles in our brains, where information was represented in temporal sequences of electrical signals, or "spikes", and thus could combine software and hardware to tackle complex problems never solved before.

"We need computers that are faster, more intelligent, and more analytical, rather than those which keep recycling old statistical or artificial intelligence methods not necessarily suited to the ... big data we're now faced with."

The largest existing neuro-computing device, called the SpiNNaker, used specially-designed ARM processors to power software that could mimic the behaviour of potentially millions to billions of human neurons.

But using Professor Kasabov's NeuCube system, researchers and industry groups could create their own data analytic systems and produce new technologies.

These could run not only on this super-computer hardware, but on other available platforms, performing much more accurately than any systems created.

At the new NeuLab, a range of projects - including eight international ones - have been based upon the NeuCube capabilities.

Professor Kasabov said data from any field - physics, neuroscience, climatology, robotics or seismology - could be applied to this technology in the pursuit of breakthroughs.

"In geophysics, people say earthquakes are not predictable; in brain science, people say we still don't understand how the brain works, we don't know how cognitive abilities can be improved or neurodegenerative diseases predicted - we've got the data but we can't analyse it effectively," he said.

"We would like to show, with small pilot experiments, that this technology can be taken further and developed for real-world solutions to complex problems."