News about the Uk Foresight Programme
Issue 1 - June 2003
Michael Kenward OBE is a freelance science and technology journalist and "writer in residence" on the Foresight Cognitive Systems Project.
Applying Brain Power to Computing
The brains in the machine - Add IT experts to brain scientists and the result could be new insights into the future of computing, says Michael Kenward
Research in computing and the brain sciences share a common language, with such terms as memory. Computers also deploy speech recognition (hearing) and visual object identification (vision). There are even attempts to emulate nose and brain to make machines with a sense of smell.
Leading lights in the physical sciences and brain sciences in the UK are looking for further synergies between these areas of science in the Foresight cognitive systems project. The exercise, has as its director Dr John Taylor, Director General of the Research Councils and previously head of HP Laboratories Bristol.
Do we really need a government initiative to get brain scientists and their computer counterparts to talk to one another? Isn't science one big endeavour with ideas bouncing around like insults at prime minister's question time? Maybe in theory, but in practice research is increasingly specialised, with huge language barriers dividing the two specialities.
IT found inspiration from the life sciences in the past. Biology provided the models for neural networks and genetic algorithms - now reasonably mature concepts in IT. But the two sides of the scientific divide drifted apart in the 1980s. The Foresight project set out to see if the time is ripe for a new round of cross fertilisation.
Two leading lights in academic research act as technical coordinators in the project. Professor Lionel Tarassenko, carries the flag for the physical sciences. The life sciences are in the hands of Professor Richard Morris of Edinburgh University.
Lionel Tarassenko, Professor of Electrical and Electronic Engineering in the Department of Engineering Science at the University of Oxford, is an expert in neural network techniques and their application to signal processing and diagnostic systems. Richard Morris, Department of Neuroscience, researches neurobiological mechanisms of memory.
"The Foresight project is a meeting of experts in both fields with a view to generating some sort of impulse to the system," says Professor Tarassenko. Professor Morris agrees. "Life scientists and physical scientists have much to learn from each other. It is an area of science in which Britain excels and has much to offer."
Various experts have written reviews of their subjects as a basis for two workshops, one for each side of the discussion. This will be followed by a major international "interaction conference" in London in September.
You don't need to look far for lessons that brains, and how they work, can bring to computing. The brain's neurons are incredibly slow by computer standards. And yet they perform feats that are beyond computers. Brains also perform with very little energy consumption - no need to fit massive cooling fans to our heads. Nor do a few dead brain cells spell the end of intelligence, thanks to biology's self-repair mechanisms.
Brains are also great at taking inputs from a broad range of sensors and combining their data - data fusion, to use the IT label. Computer designers would love to steal some of these tricks. Can IT researchers find a silicon and software equivalent of the ability forget, to clear out useless information?
It isn't just the hardware that interests IT specialists. Our brains, well some of them, have some pretty nifty software, working with incomplete information, making decisions according to the environment, like noticing immediately when some mutters your name across a crowded room.
Such technologies as speech recognition may have come a long way in the past decade, but it is all done with statistics and complicated mathematical models. The brain could teach us a lot here.
Dr Taylor uses another example of where IT could use some input from brain research. programming. "In future you may have to educate your computer rather than program it," he says. "Then you have to ask yourself, how do you test it to see that it does what you want? Does it have to sit an examination?"
This summer, Foresight's web site will open up to wider view a series of documents on these and other aspects of IT and the brain sciences. The idea of this exercise, and of the meetings of scientists and experts from the IT business, is to home in on some key questions that researchers would love to get their teeth into, and where there might be benefits from a union of different areas of science.
The hope is that ideas will flow in both directions. After all, IT researchers have achieved a lot with their miserably inflexible hardware. Maybe research in artificial neurons, for example, and how they work could throw some light on the workings of the brain. Then there are mathematical concepts that can help the lifer sciences. Imaging techniques from the physical sciences also allow us to study the working brain. But the most profitable cross fertilisation could come in ways that we have yet to imagine, which is where the Foresight initiative hopes to play a part.
Other links