ECAI 2000 Workshop on

Machine Learning in Computer Vision

Tuesday, 22nd August 2000

before
14th Biennial European Conference on Artificial Intelligence (ECAI-2000)
Humboldt University, Berlin

under the auspices of
MLnet II - the European Network of Excellence in Machine Learning

Technical Description

Learning is one of the next challenging frontiers for computer vision research, and it has been receiving increasing attention in the recent years. This workshop will provide a forum for discussing current research in AI and pattern recognition that pertains to machine learning in computer vision systems.
From the standpoint of computer vision systems, machine learning can offer effective methods for automating the acquisition of visual models, adapting task parameters and representation, transforming signals to symbols, building trainable image processing systems, focusing attention on target object. To develop successful applications, however, we need to address the following issues:

How is machine learning used in current computer vision systems?
What are the models of a computer vision system that might be learned rather than hand-crafted by the designer?
What machine learning paradigms and strategies are appropriate to the computer vision domain?
How do we represent visual information?
How does machine learning help to transfer the experience gained in creating a vision system in one application domain to a vision system for another domain?

From the standpoint of machine learning systems, computer vision can provide interesting and challenging problems. Many studies in machine learning assume that a careful trainer provides internal representations of the observed environment, thus paying little attention to the problems of perception. Unfortunately, this assumption leads to the development of brittle systems with noisy, excessively detailed or quite coarse descriptions of the perceived environment. Some specific machine learning research issues raised by the computer vision domain are:

How dealing with noisy observations?
How can large sets of images with no annotation be used for learning?
How dealing with mutual dependency of visual concepts?
What are the criteria for evaluating the quality of learning processes in computer vision systems?
When a computer vision system should start/stop the learning process and/or revise acquired models?
When is it useful to adopt several representations of the perceived environment with different levels of abstraction?

The workshop will maintain a balance between theoretical issues and descriptions of implemented systems to promote synergy between theory and practice. Work in areas such as statistical pattern recognition is also welcome.

Topics

The workshop will maintain a balance between theoretical issues and descriptions of implemented systems to promote synergy between theory and practice. Works in areas such as statistical pattern recognition are also welcome. Topics of interest include, but are not limited to:

Learning to recognize shapes
Supervised learning of visual models
Unsupervised learning for structure detection in images
Multistrategy learning in vision
Learning and refining visual models
Multi-level learning and reuse of learned concepts
Learning important features for image analysis
Relational learning in vision
Context in visual learning
Mining from large collections of images and videos
Interpretation of discovered visual models
Image segmentation via learning
Probabilistic model estimation and selection
Applications such as medical imaging, object recognition, remote sensing, digital maps, document image analysis and recognition, spatial reasoning

Workshop structure and attendance

The workshop is aimed to be a high communicative meeting place for researchers working on similar topics, but from different communities. In order to achieve these goals, workshop will consist of one or two invited talks, followed by short presentations and longer discussions. Each author will be encouraged to read another accepted paper and to comment on it after the original talk was given.
Workshop attendance will be limited to registered participants. Workshop delegates MUST register for the main conference. Hence, registration for the workshop is through the main registration for ECAI 2000. To ensure availability of places at a workshop, delegates should register as soon as possible.

Submission Procedure

Authors are invited to submit original research contributions or experience reports in English. Submitted papers must be unpublished and substantially different from papers under review. Papers that have been or will be presented at small workshops/symposia whose proceedings are available only to the attendees may be submitted.

Papers should be double-spaced and no longer than 12 pages. Papers should be sent electronically (postscript or pdf) not later than March 20. 2000 to

Donato Malerba <malerba@di.uniba.it>

Papers will be selected on the basis of review of full paper contributions. Authors should make certain that the learning techniques they describe address the special issues that are associated with problems in computer vision.

Final camera-ready copies of accepted papers will be due by June 01 2000. The proceedings will be distributed at the workshop.

Extended versions of the papers are invited for the a special issue of the journal Applied Artificial Intelligence (Call for paper)

Style Guide

The style for papers in the workshop notes is the same as for ECAI proceedings.

It is described in an example paper available as a postscript document and a PDF document. It is also supported through the provision of latex style files, including the source for the guidelines paper.

Important Dates

Deadline for papers:	April 10, 2000
Notification of acceptance:	May 01, 2000
Camera-ready copies of papers:	June 01, 2000
Workshop on ECAI-98:	August 22, 2000

Organizing Committee

This workshop will be organized by the following people:

Joachim M. Buhmann, University of Bonn, Germany
Terry Caelli, The University of Alberta, Alberta, Canada
Floriana Esposito, University of Bari, Italy (cochair)
Donato Malerba, University of Bari, Italy (cochair)
Petra Perner, Institute of Computer Vision and Applied Computer Science, Leipzig, Germany
Maria Petrou, University of Surrey, UK
Tomaso A. Poggio, MIT, Boston, MA
Alessandro Verri, University of Genoa, Italy
Tatjana Zrimec, University of Ljubljana, Slovenia

Travel support

MLNET members can apply for travel support if they contribute to the workshop. Information on travel support can be found at http://www.mlnet.org/mlnet2/services.html.

Donato Malerba

Last modified: Wed June 14 15:11:31 DFT 2000