ICML-2002 Workshop
on
Development of Representations

July 9th, 2002

Contents and related links:

Workshop Description

Online Proceedings

Workshop schedule

Call for Papers

ICML-2002 main page

Workshop Description

The representation of a learning problem has long been known to be a major factor in learning performance. The nature of appropriate representations and representational change as a part of the learning process have been studied in a variety of forms in a number of subfields within machine learning, artificial intelligence and, more recently, other communities. Despite this fact, representations are typically hand-coded rather than acquired automatically. The goal of this workshop is to explore problems in the area of automated development of representations and to build ties between the various relevant communities.

From a pragmatic standpoint, many learning problems of interest involve extremely large search spaces when viewed in terms of their basic input features. Examples include learning useful behavior for a robot that receives a continuous stream of video input, or learning to play the game of Go. For such problems, an unbiased search is infeasible, and a bias must be employed that focuses the search within the input space so that the size of the problem is effectively reduced. Letting representations develop as part of learning may be viewed as a way of establishing such a bias.

Online Proceedings

Marco Botta, Attilio Giordana, and Paolo Terenziani.
Discovering Complex Events in Long Sequences.

Edwin D. de Jong and Tim Oates.
A Coevolutionary Approach to Representation Development.

Sarah Finney, Natalia H. Gardiol, Leslie Pack Kaelbling, and Tim Oates.
The Thing That We Tried Didn't Work Very Well: Deictic Representation in Reinforcement Learning.

Jeffrey Forbes and David Andre.
Representations for Learning Control Policies.

Elias Gyftodimos and Peter A. Flach.
Hierarchical Bayesian Networks: A Probabilistic Reasoning Model for Structured Domains.

Bernhard Hengst.
Discovering Hierarchy in Reinforcement Learning with HEXQ.

Alberto Paccanaro.
Learning Distributed Representations of Concepts from Relational Data.

Leonid Peshkin and Edwin D. de Jong.
Context-based policy search: transfer of experience across problems.

Martijn van Otterlo.
Relational Representations in Reinforcement Learning: Review and Open Problems.

Alexei Vinokourov, John Shawe-Taylor, and Nello Cristianini.
Finding Language-Independent Semantic Representation of Text Using Kernel Canonical Correlation Analysis.

Workshop Schedule

----------------
9:00 9:15 Opening remarks
Edwin de Jong and Tim Oates
9:15 10:15 Invited Talk: Many-Layered Learning
Paul Utgoff
10:15 10:35 Finding Language-Independent Semantic Representation of Text Using Kernel Canonical Correlation Analysis
Alexei Vinokourov, John Shawe-Taylor, and Nello Cristianini
10:35 11:00 Break
11:00 11:20 Learning Distributed Representations of Concepts from Relational Data
Alberto Paccanaro
11:20 11:40 Relational Representations in Reinforcement Learning: Review and Open Problems
Martijn van Otterlo
11:40 12:00 The Thing That We Tried Didn't Work Very Well: Deictic Representation In Reinforcement Learning
Sarah Finney, Natalia H. Gardiol, Leslie Pack Kaelbling, and Tim Oates
12:00 13:30 Lunch
13:30 13:50 Context-Based Policy Search: Transfer of Experience Across Problems
Leonid Peshkin and Edwin D. de Jong
13:50 14:10 Representations for Learning Control Policies
Jeffrey Forbes and David Andre
14:10 14:30 Discovering Complex Events in Long Sequences
Marco Botta, Attilio Giordana, and Paolo Terenziani
14:30 14:45 Break
14:45 15:05 A Coevolutionary Approach to Representation Development
Edwin D. de Jong and Tim Oates
15:05 15:25 Hierarchical Bayesian Networks: A Probabilistic Reasoning Model for Structured Domains
Elias Gyftodimos and Peter A. Flach
15:25 15:45 Discovering Hierarchy in Reinforcement Learning with HEXQ
Bernhard Hengst
15:45 17:00 Group discussion

Program co-chairs

Edwin de JongBrandeis University
Tim OatesUniversity of Maryland Baltimore County

Program committee

Jonathan BaxterWhizBang! Labs
Rich CaruanaCornell
Rod GrupenUniversity of Massachusetts, Amherst
Tom HeskesUniversity of Nijmegen, The Netherlands
Leslie KaelblingMIT
Justus PiaterINRIA Rhone-Alpes, France
Jude ShavlikUniversity of Wisconsin, Madison
Paul UtgoffUniversity of Massachusetts, Amherst