Inference methods based on graphical structures of knowledge

Inference methods based on graphical structures of knowledge

WIGSK'06

A one day ECAI workshop, Riva del Garda, Italy, 29th August 2006.
www.irit.fr/LC/WIGSK06

The use of graphical structures to improve the efficiency of inference has been intensively developed in the last few years. In particular, local computation on join trees (including variable/bucket elimination) has proved to be a very general method of computation in reasoning formalisms, and for knowledge compilation. It has been shown to be applicable, in particular, to probability, possibility theory and other uncertainty calculi, relational databases, constraints including finite constraints, soft constraints, linear constraints and inequalities, and a wide range of logics, including propositional and first-order classical logics and modal logics.

Recently it has also be shown that another family of representations can be applied for many of the same problems; this family includes NNFs, tree-automata, AND/OR graph representations and decision diagrams. In addition, incomplete graph-based propagation algorithms have been used in the solving of hard and soft constraints.

The focus of this workshop will be on these graphical approaches, and their application to uncertain inference, reasoning with soft constraints and preferences, decision making and related problems.

This workshop is a follow-up of the ECAI-04 workshop on Local Computation for Logics and Uncertainty.

Please note that workshop attendees are required to register also for the main conference.
See the ECAI 2006 registration page.

PROGRAMME

8.50 Welcome and introduction

9.00 Invited talk: Cedric Pralet
Structuring Multi-operator Variable Eliminations on Graphical Models: Towards Multi-operator Cluster Trees and Multi-operator Cluster Dags

10.00 - 10.30 coffee break

10.30 Philippe Jégou, Samba Ndojh Ndiaye and Cyril Terrioux
Strategies and Heuristics for Exploiting Tree-decompositions of Constraint Networks (download pdf)

11.00 Le Chang and Alan K. Mackworth
Constraint-Based Inference Using Local Consistency in Junction Graphs (download pdf)

11.30 Christian Eichenberger
Local Computation with Gaussian Potentials (download pdf)

12.00 Marc Pouly
Minimizing Communication Costs of Distributed Local Computation (download pdf)

12.30 - 2.00 lunch

2.00 Invited talk: Rina Dechter
Advanced Reasoning in Graphical Models

3.00 Michael Wachter
Representing Boolean Functions with Propositional Directed Acyclic Graphs (download pdf)

3.30 - 4.00 coffee break

4.00 Cesar Schneuwly
Updating in Local Computation (download pdf)

4.30 Further discussion

ORGANIZING COMMITTEE

Adnan Darwiche
Computer Science Department, University of California, Los Angeles, USA

Rina Dechter.
School of Information and Computer Science, University of California, Irvine, USA

Hélène Fargier
Institut de Recherche en Informatique de Toulouse, Université Paul Sabatier, Toulouse, France.

Jürg Kohlas
Department of Informatics, University of Fribourg, Switzerland.

Jérôme Mengin
Institut de Recherche en Informatique de Toulouse, Université Paul Sabatier, Toulouse, France.

Gérard Verfaillie
ONERA, Toulouse, France.

Nic Wilson
Cork Constraint Computation Centre, University College Cork, Ireland.

Author: Jérôme Mengin
Last modified: August 29, 2006