For post-tutorial information
please see the
'AFTER TUTORIAL' web site.
REPRESENTATION AND LEARNING IN ROBOTS AND ANIMALS
A two-day tutorial at
University of Edinburgh George Square
on behalf of the
EC-Funded CoSy Project
Date for tutorial: Saturday 30th July and Sunday 31st July, 2005
Dates for IJCAI: 30th July -- 5th August 2005
We gratefully acknowledge sponsorship by
BOOKLET FOR TUTORIAL
The tutorial was held in the
David Hume Tower
Faculty Room North,
George Square, EH8 9JX
on AI-Inspired Biology (AIIB)
(Not Biologically Inspired AI -- BIAI!)
Past, ongoing and especially future influences from AI/robotics and
artificial cognition to the study of natural cognition.
at AISB2010 the 36th Annual
Convention of SSAISB (Funded by EUCogII)
MeetingOfMinds Workshop, Paris, Sept 2007.
3-4 April 2006:
AISB'06 GC-5 symposium at AISB'06 in Bristol
euCOGNITION: European Network for the Advancement of Artificial Cognitive Systems.
28 Jul 2005: Revised title for Jackie Chappell's talk in Session 4
See Programme below. Abstract in
14 Jul 2005: Changed plans for session 4, day 1
Unfortunately Lesley Kaelbling is now unable to attend the tutorial as
originally planned. So plans for session 4 (16:00-17:30) have been
Dr Jackie Chappell
(The University of Birmingham, UK) and
Professor Ales Leonardis
(University of Ljubljana,
Slovenia) have agreed to give presentations
instead. For details see
the tutorial schedule.
25 Jun 2005: New online map utility
If you are planning travel I recommend google's
new map and route finding web site.
To find the location of the tutorial (Appleton Tower)
type this in the search panel: Edinburgh EH8 9LE
You can drag the map around continuously, and zoom in and out.
The main IJCAI
conference centre (EICC)
at EH3 8EE, entrance on
18 Jun 2005:
The schedule for day 1 has been rearranged on account of constraints on
travel of speakers.
7 Jun 2005:
Unfortunately Alex Kacelnik cannot stay for the second day
owing to family commitments.
Fortunately, his former colleague
Dr. Jackie Chappell
at the University of Birmingham
will be at the tutorial on both days and has agreed to join the panel on
20 May: ONE DAY LEFT before the
IJCAI tutorial and conference registration fees
24 March: Tutorial programme updated
24 March: Registration details, fees, etc. online
A two day advanced tutorial supported in part by
the European Commission's
Cognitive Systems initiative, at which tutorials will be presented
by seven leading researchers on problems of learning and
representation in integrated natural and artificial systems
combining multiple functions using different forms of
representation and learning. The presentations will illustrate the
state of the art and unsolved problems.
The tutorial will end with a panel discussion on what the
major unsolved problems are and how we can make progress towards
solving them. This is part of the
This two-day tutorial arises out of the European Commission's 6th
It is organised by two members (Bernt Schiele and Aaron Sloman) of one
of the recently funded integrated projects in that initiative,
CoSy, described here:
as part of that project.
The tutorial is organised as a collection of 90 minute sessions, in
which presentations are given by leading researchers, who will cover a
variety of topics in representation and learning, both in artificial
systems and in biological systems, including a tutorial on animal
cognition, and surveys of important unsolved problems.
There should be opportunities for discussion throughout the two days,
both during and between the formal sessions. In adddition,
the final session on the second day
will consist of a panel discussion involving
all the presenters and members of the audience, attempting to identify
the most important unsolved problems and good ways of addressing them.
The overall aim is to provide attendees with an up to date survey of
major points of theory and practice in the design and implementation of
complete systems using multiple forms of representation and multiple
forms of learning, for instance in robots, or intelligent control
systems combining perception, learning, reasoning, planning, action
(including 3-D manipulation) problem-solving, language understanding and
self-understanding. The focus is on gaps and inadequacies in our current
understanding, as well as achievements.
A brief presentation on some important gaps in our perception and
understanding of 3-D shape and affordances can be found
The Cognitive Systems initiative within the European Community's
Sixth Framework Programme (FP6) described here
in a presentation by Colette Maloney in October 2004, is a visionary
research programme sponsoring long term research on very difficult
questions about intelligent integrated systems combining many aspects of
human competence. The emphasis is on science rather than applications,
though the research has many potential applications. The 8 projects in
the first wave funded by this initiative, starting in
September 2004, were summarised
in this presentation
by David Vernon at The Hague,
in November 2004.
One of the requirements for the funded projects is to engage in training
of researchers working on the projects. Another objective is 'community
building'. This advanced tutorial aims to meet the normal objectives of
IJCAI tutorials and, in addition, it aims to:
provide state of the art presentations by internationally leading
researchers especially for the benefit of younger researchers working on
Cognitive Systems, with a special focus on problems of representation
and learning in the design of systems that integrate multiple functions,
such as visual and auditory perception, speech understanding and
generation, various kinds of learning and development, planning and plan
execution, self-understanding, social interactions;
introduce lessons from a number of disciplines including AI, Cognitive
Science, Linguistics, Psychology, Neuroscience, Animal Behaviour
studies and Philosophy.
help to build a community of researchers thinking and communicating
about these problems within Europe and further abroad, by using the
opportunity provided by a major international AI conference located in
Edinburgh, itself a major European AI centre,
identify opportunities for subsequent collaborative activities involving
a broad-based international research community focusing on the problems
being addressed by the EU projects, including collaboration
with researchers on related national projects, such as the DARPA and NSF
Cognitive Systems initiatives and the Foresight Cognitive Systems
initiative and the Computing Research Grand Challenges in the UK --
publicise the existence of the EU Cognitive Systems projects in order to
stimulate approaches from outstanding researchers who may be interested
in collaborating in our work.
People with common interests who first meet at the tutorial will have
opportunities throughout the rest of the conference (which ends on 5th
August) to to get to know one another and plan
The theme of the tutorial is closely related to one of the Computing
Research Grand Challenges sponsored by the UKCRC, outlined in this
Grand Challenges in
edited by Tony Hoare and Robin Milner. One of the seven challenges is
GC5: 'Architecture of Brain and Mind'.
OUTLINE OF THE TUTORIAL
The two days will be organised in 2 hour sessions, covering a variety of
distinct types of learning involving different sorts of representation,
including hybrid systems in which there are strong interactions between
subsystems. This could include, for instance, both symbolic and
sub-symbolic mechanisms, statistical and non-statistical learning,
learning practical skills and learning new forms of self-understanding,
along with interactions between very different parts of a complex
architecture, such as reactive, deliberative and reflective components.
A typical integrated system could be an animal, or child, or robot in
which many capabilities are combined. A tutorial on animal cognition
from a leading researcher is included. There will also be links to human
cognition, and recent work on brain imaging.
The primary aim will be to deepen scientific understanding of both the
problems of designing integrated artificial systems combining multiple
capabilities and the problems of explaining and modelling naturally
occurring integrated systems.
SUMMARY PROGRAMME (NB: SUBJECT TO CHANGE)
Day 1: Saturday 30th July
SESSION 1 (09:00--10:30):
Talk 1: Animal Learning, Machine Learning
SESSION 2 (11:30--12:30):
Talk 2: Animal Learning, Representation and Choice.
SESSION 3 (14:00--15:30):
Talk 3: Architectures for Cognitive Information Processing
Jackie Chappell and
TALK 4a (16:00--16:45): Jackie Chappell will talk on:
How do animals gather useful information about their
environment and act on it?
TALK 4b (16:45--17:30): Ales Leonardis will talk on
Problems of representation
learning in machine vision.
Day 2: Sunday 31st July
SESSION 5 (09:00--10:30):
Talk 5: Words and pictures
SESSION 6 (11:00--12:30):
Talk 6: Planning and Learning in Hybrid Discrete-Continuous Models
SESSION 7 (14:00--15:30):
Talk 7: Plans and the Computational Structure of Language
SESSION 8 (16:00--17:30):
Final session: Discussion between speakers and audience,
led by the tutorial organisers.
Alex Kacelnik will not be available for this. His place will be taken by
Jackie Chappell, his former colleague.
A booklet of notes by the organisers and presenters will be available at the
The tutorial organisers will also lead the final panel session on the
second day, involving all the presenters.
POTENTIAL TARGET AUDIENCE
The tutorial could be useful for PhD students and other researchers in
all areas of AI and Cognitive Science who wish to broaden their
understanding of current problems and approaches in the study of
representation and learning in a variety of contexts. The tutorial
should be especially useful for those who wish to understand how to
design robots and synthetic agents
in which different kinds of functionality are integrated.
It should also provide opportunities to get to know other researchers
who are interested in building bridges between topics that are normally
There will be no specific prerequisites apart from an interest in
AI and/or Cognitive science.
GENERAL REQUIREMENTS FOR IJCAI TUTORIALS
Motivate and explain a topic of emerging importance for AI
The tutorial should help attendees gain a new understanding
of problems of integration of diverse capabilities.
Survey a mature area of AI research and/or practice
Some parts of the tutorial will survey established
learning techniques, their scope and their limitations,
along with established powerful forms of representation and
Present a novel synthesis combining distinct lines of AI work
The tutorial will juxtapose a variety of types of problems
and techniques that involve learning using different forms of
representation in different parts of a complex architecture. This
will enable participants to encounter new ways of thinking about
problems of integration in 'complete' intelligent systems.
Registration, Fees, Accommodation and Travel
Further details regarding fees, registration and accommodation
can be found on the
IJCAI-05 Web Site
Fees for the tutorial vary according to whether you are registered for
the main conference or not, whether you are a student on 1st Aug 2005 or
not, whether you register before 21 May or not.
Most IJCAI tutorials last only half a day. This one lasts two days, and
is therefore the equivalent of four ordinary tutorials. However
the registration fee for this tutorial is only twice the normal tutorial
fee because no funding for speakers at this tutorial
is provided by IJCAI, and
only one tutorial booklet will be produced insteaad of four.
Fee information as on online registration form on 31 Mar 2005 (all
inclusive of VAT at 17.5%):
|Early Tutorial Fee
||(prior to/on 21 May 2005)
|Late Tutorial Fee
||(after 21 May 2005)
|On-site Tutorial Fee
|Student Tutorial Fees
|Early Student Tutorial Fee
||(prior to/on 21 May 2005)
|Late Student Tutorial Fee
||(after 21 May 2005)
|On-site Student Tutorial Fee
|Non-delegate Tutorial Fees
|Tutorial Only Fee
|Student Tutorial Only Fee
The Two-day tutorial costs twice the applicable half-day tutorial
Last updated: 5 Sep 2005
School of Computer Science,
The University of