ECML-93 Workshop on Integrated Learning Architectures
The Integrated Learning Architectures Workshop, ILA-93 was organized by
the European Conference on Machine Learning held in Vienna, April 8
1993.
A discussion on the motivation and relevance for this workshop, authored
by the Program Committee may be foun here at the
Integrated Learning Architectures article.
Description
The goal of the ILA-93 workshop was to bring together researchers who are
actively building integrated systems that involve learning tasks. The
workshop focus was on exixting and proposed learning architectures that
integrate multiple learning methods or that integrate learning with the
performance aspect of the architecture, as well as the problems that are
inherent to creating integrated learning architectures. The performace
aspect may deal with any kind of problem solving task such as recognition,
planning, or model-based reasoning. The emphasis of the workshop was on
the integration issues rather than on the learning per se: How do learning
and proplem solving constrain each other? How can they support each other?
How do learning processes interact with each other in the context of the
problem solving of a specific architecture?
Different approaches, however, have to have some common grounds to be
considered integrated learning architectures. To state clearly the
constraints for workshop scope, three conditions have been defined for
ILAs:
- some learning has to be involved in the system,
- learning has to be integrated in the system (and not merely used to
construct the system), and
- the performance component of the architecture has to be well specified.
The main point here is distinguishing learning embedded in a global
system from mere usage of learning methods instrumental to (but external
to ) build, say, an expert system. The workshop scope was circumscribed to
improve understanding of learning's role into a wider system, be it an
intelligent agent or a task-specific architecture.
Workshop Papers
The following papers were reviewed and accepted by the Program Committee
- A Multistrategy Knowledge Refinement and Acquisition Toolbox,
Nicolas Graner, D. Sleeman,
graner@csd.abdn.ac.uk
- New Algorithms for ALEX: Expanding An Integrated Learning,
Lothar Winkelbauer, Petr Berka,
lothar@iiasa.ac.at, berka@ai.univie.ac.at
- Using reflection principles in the integration of learning and
problem solving,
E Plaza & J L Arcos,
arcos@iiia.csic.es
- Integrating Abstraction, Explanatyion-based Learning from
Multiple Examples, and Hierarchical Clustering with a Performance
Component for Planning,
Ralph Bergmann,
bergmann@informatik.uni.kl.de
- A Method of Symbolic Machine Learning based on a Geometrical
Description Language. The Fongus System: Application to the game of
Go,
Pierre Pompidor,
pompidor@lirmm.lirmm.fr
- Induction and Reasoning from Cases,
Michel Manago, Klaus-Dieter Althoff, Ralph Traphoner,
althoff@informatik.uni-kl.de
- Using event calculus to integrate planning and learning into an
intelligent autonomous agent,
G Sablon and M Bruynooghe,
gunther.sablon@cs.kileuven.ac.be
- A developmental approach for integrated architectures,
M Martin & U Cortes,
mmartin@lsi.upc.es
Program Committee
- Agnar Aamodt, University of Trondheim,
agnar@ifi.unido.no
- Enric Plaza, IIIA, CSIC, enric@iiia.csic.es
(coordination)
- Ashwin Ram, Georgia Institute of Technology,
ashwin@cc.gatech.edu
- Walter van de Velde, AI-Lab, Vrije Universiteit Brussels,
walter@arti.vub.ac.be
- Maatten van Someren, Universiteit van Amsterdam,
maarten@swi.psy.uva.nl