Fighting spam and thinking robots – bio-inspired AI


6 Aug 2008

Share on FacebookTweet about this on TwitterShare on LinkedInShare on Google+Pin on PinterestShare on RedditEmail this to someone

Share on FacebookTweet about this on TwitterShare on LinkedInShare on Google+Pin on PinterestShare on RedditEmail this to someone

Can anti-spam software learn to distinguish between harmful and harmless email? Will robots ever be able to win in a spontaneous dance-off? These questions and their answers are inextricably linked to modelling artificial intelligence on bio systems found in nature.

This week’s Artificial Life (ALIFE XI) conference held at the University of Southampton in the UK will feature lectures that tackle these issues and will introduce some of the most complex and interesting developments in robotics to date.

Artificial intelligence is about getting machines to think and learn in a plastic sense like the human brain and body – if spam software can adapt like the human immune system, then it has a better chance of fighting off rogue emails.

How do we fight the ever-evolving complexity and increasing amount of spam arriving in our inboxes by the hundreds on a daily basis? As soon as keywords are identified and used to eliminate certain waves of spam, the spammers adapt and we’re back to square one.

Then there is the problem of legitimate mail getting trapped in the spam folder if they themselves trigger keywords.

Alaa Abi-Haidar and Luis Rocha from the department of informatics, Indiana University, Bloomington, USA and the Instituto Gulbenkian de Ciencia, Portugal respectively, believe their spam detection model based on the immune system is the answer.

The two researchers will explain how their spam detection system is modelled on the vertebrate immune system in that it learns to detect between harmless and harmful antigens – a system which they claim is as effective as state-of-the-art spam binary classifiers.

Fighting spam aside, what about real live robots? A small, egg-shaped robot from Japan called Miuro (pictured) dances to music from its partial-iPod brain by using complex mathematics to come up with the perfect boogie.

Although previous Sony robots have danced elegantly to both dance and ceremonial Japanese music, they have been programmed and would continue to go through the motions whether the music played or not.

Miuro moves about based on its calculations of rhythm extracted from the music played – in other words it does not dance anything like your uncle at the last family wedding and can avoid obstacles in its path.

One robot on display at ALIFE that has immediate scientific applications is Roke’s DORA (demonstration of robot autonomy). DORA maps, learns and navigates all at once, literally learning from scratch how to find its way around new terrain.

While this is all based on vision processing, DORA can construct a 3D environment based on objects in its surrounding area.

Funnily enough, MIT in the US has worked on a ‘blind’ robot that maps its environment out using sonar. I wonder if anyone has created the sonar goggles that Batman wears in The Dark Knight?

By Marie Boran

Pictured:Miuro the dancing robot