Brain-training to spark granular computing

 

Gallery

By Stewart MItchell

Posted on 13 Dec 2008 at 14:36

A human approach to processing raw data could revolutionise the way computers deal with complex information such as weather modelling, according to academics at the University of Nottingham.

Researchers in the School of Computer Science at the University's Malaysia Campus are exploring "granular computing" - a computer paradigm that processes sets of information rather than individual bits in the way data is currently managed.

By looking at data in this way, new patterns and relationships emerge and they could herald new types of computer modelling in a range of fields, including process control and optimisation, resource scheduling, climate change and bioinformatics.

The concept of a granular approach to computing is inspired by human thought processes, says Professor Andrzej Bargiela, director of computer science at Malaysia Campus.

"Creating abstractions from detailed information is essential to human knowledge, interaction and reasoning," says Bargiela. "The human brain filters the flood of information and distils knowledge subconsciously.

"We remember conversations, but we don't remember every word - the raw data - we remember the meaning, gist and nuance, the abstractions of the conversation. If we were aware of every single thing, our minds would be overloaded."

The researchers claim the granular computing approach to information processing could capture this essential characteristic of human information processing and offer a breakthrough in dealing with information overload in all sorts of computers used for weather modelling, traffic monitoring, job scheduling, timetabling and protein classification.