Leeds' entry wins BCS Machine Intelligence Prize

The BCS Machine Intelligence Prize 2004 was won by a cognitive vision system developed in the School of Computing at the University of Leeds. The system was demonstrated live at AI-2004 in the final and voted on by the conference audience. The prize is awarded for demonstration of progress towards Machine Intelligence and is sponsored by Electrolux. Leeds' winning entry was a product of the COGVIS research project, in which perceptual learning is integrated with task induction.

Live Demo	Audience Participation	Prize Presentation

Motivation

Can we build an artificial system that can observe events in an unknown scene, with unknown objects, and learn to participate in a similar manner to a child?

Integrating perceptual learning with task induction

In this work, our overall aim is to point a camera at a scene, and learn to particpate in the activity. This is illustrated by the computer learning how to play the game "paper, scissors, stone". First the world is observed in a training phase. Models are learned, then the computer is able to play the part of a participant, interacting with the world. (Click on the image to play the movie).

Training Phase	Demo Playing Phase

Entry Description

A cognitive vision system for autonomous learning from observation is presented. The system has been trained on a sequence of two players playing a card game. An attention mechanism is used to segment the continuous input audio-visual streams, from which features are extracted and clustered in an unsupervised manner. From this, a symbolic data stream is formed, representing the perceived state of the world (what's on the table) and the actions (speech of the participants). Inductive Logic Programming is used to generalise a set of hypotheses from the examples, which in turn are used by an inference engine to drive a talking head which can interact with the game. It is this synthetic agent with its learned perceptual classes and protocol rules which will be demonstrated on live video input. Link to movies.

Contributors and funding

This demonstration was produced by Chris Needham, Derek Magee, Paulo Santos, Vincent Devin, David Hogg and Tony Cohn, in the School of Computing at the University of Leeds. This research was undertaken as part of the CogVis project (Cognitive Vision) which was a European Union funded collaborative project (Contract IST-2000-29375) to to study the design of Cognitive Vision Systems.

MSc Cognitive Systems

NEW for 2005 ENTRY: Masters degree in Cognitive Systems.

This programme offers a broad grounding in cognitive systems, as preparation for a career using advanced IT skills and knowledge, or for further research. Modules reflect the School's research in systems thinking, modelling, envisioning information, machine learning, vision, language, reasoning and bioinspired computing. For more information see the MSc Cognitive Systems Page.

Press Releases and Articles

24 January 2005
New Scientist Article

Machine learns games 'like a human'

22 December 2004
PRESS RELEASE Issued by Electrolux

Artificial intelligence in machines is child’s play

Contact Details

email:        chrisn@comp.leeds.ac.uk
tel:             +44 113 343 5767
fax:            +44 113 343 5468
room:         7.14 E C Stoner
address:    School of Computing, The University of Leeds, Leeds, LS2 9JT