Parallel and Distributed computing for Machine Learning


In conjunction with the
14th European Conference on Machine Learning (ECML'03)
7th European Conference on Principles and Practice of Knowledge Discovery in Databases (PKDD'03)


22-26 September 2003
Cavtat-Dubrovnik, Croatia

ECML/PKDD-2003


Workshop Proceedings in PDF format
The invited talk slides in PDF format


Workshop Topics and Goals
Invited speakers
Tentative Workshop Program
Format
Intended Audience and Participants
Paper Submission Guidelines
Important Dates
Organization
Sponsor

First invited speaker
Professor Domenico Talia
DEIS
Universitá della Calabria

Professor Domenico Talia will be the invited speaker of the workshop. Professor Talia has extensive experience in both parallel and distributed data mining and on Grid-based knowledge discovery. More information can be found at his home page.

Second invited speaker
Professor Hillol Kargupta
Department of Computer Science and Electrical Engineering
University of Maryland Baltimore County

Professor Hillol Kargupta will be the workshop second invited speaker. Professor Kargupta has extensive experience in distributed Data Mining. More information can be found at his home page .


Workshop Goals and Topics

Parallel and distributed computing is of most importance for Machine Learning (ML) practitioners. Taking advantage of a parallel or a distributed execution a ML system may: i) increase its speed; ii) search a larger space and reach a better solution or; iii) increase the range of applications where it can be used (because it can process more data, for example). The development of fast ML systems capable of processing large amounts of data is quite a relevant issue for both ML researchers and KDD practitioners using ML systems.

The workshop will be concerned with the exchange of experience among researchers that use parallel or distributed computing within ML. Researchers will present recently developed algorithms/systems, on going work and applications taking advantage of such parallel or distributed environment.

The topics of interest include (but are not restricted to) the following ones:

Tentative Workshop Program

The TOC of the proceedings in PDF format.

Talks will be 20 minutes long with 5 minutes for questions.

Session I (9:00 - 10:30)
9:00 - 9:05
Opening
9:05 - 10:05
Invited talk: Parallel and Distributed Data Mining: from Multicomputers to Grids
the slides of the talk
Prof. Domenico Talia
10:05 - 10:30
A New Method for Combining Partitions, Applications for Cluster Ensembles in KDD
Pierre-Emmanuel JOUVE, Nicolas NICOLOYANNIS LABORATOIRE ERIC
10:30- 11:00 Coffe Break

Session II (11:00 - 12:40)
11:00 - 11:25
Context-based Distributed Regression in Virtual Organizations
Yan Xing, Michael G. Madden, Jim Duggan, and Gerard J. Lyons
11:25 - 11:50
Comparing the Parallel Automatic Composition of Inductive Applications with Stacking Methods
Hidenao Abe, and Takahira Yamaguchi
11:50 - 12:40
Invited talk: Privacy Sensitive Distributed Data Mining from Multi-Party Resources
Prof. Hillol Kargupta
12:30- 2:00 Lunch Break

Session III (2:00 - 3:30)
2:00 - 2:25
Multi-way Distributed SVM algorithms
Francois Poulet
2:25 - 2:50
A parallel ILP algorithm that incorporates incremental batch learning
Nuno Fonseca, Rui Camacho and Fernando Silva
2:50 - 3:15
A Data-Parallel Version of Aleph
Stasinos Th. Konstantopoulos
3:15 - 3:40
Exploiting Parallelism in Decision Tree Induction
Nuno Amado, Joao Gama, Fernando Silva

Format

The workshop will include an invited talk. The invited talk would be given by a researcher with experience in parallel or distributed execution of ML systems. He will report on acquired experience on the domain and point out research directions. The rest of the workshop schedule would include the research papers presentations and a final discussion panel. The discussion panel will involve the invited speaker and researchers from the organizing committee. Each person in the panel will give (in less than 5 minutes) his opinion on future directions and application for research. This should suggest question and lead to the involvement of the attending audience.

Intended audience and participants

The workshop research topic will be of direct interest for almost all ML and KDD practitioners and therefore for the researchers attending ECML-2003 and PKDD-2003. The workshop could also be of interest for researchers in parallel or in distributed computing that may use ML and KDD applications as testbeds for their systems. Attendance is not limited to the paper authors.

Paper Submission Guidelines

Authors are invited to submit papers addressing one or more of the topics presented above. Papers should be prepared according to ECML/PKDD-2003 Instructions for Authors, and should not exceed 12 pages. Acceptable formats are PostScript or PDF.

Please send the papers by e-mail to
[email protected], cc: [email protected]
Subject: ParallelDistributedML-2003 workshop submission paper.

Submitted papers will be reviewed by referees from the Program Committee. Accepted papers will be published in the working notes provided by ECML/PKDD-2003 and possibly the selected papers will be published as a Springer series book. All accepted papers will be published in the workshop Web page.

Important Dates

Paper submission deadline June 13, 2003
Paper acceptance notification July 4, 2003
Paper camera-ready deadline July 11, 2003
Workshop at ECML/PKDD-2003 September 23, 2003



Organization



Program Chairs

Rui Camacho
LIACC and FEUP, Universidade do Porto, Portugal
[email protected]


Ashwin Srinivasan
Oxford University Computing Laboratory
[email protected]

Program Committee

Hendrik Blockeel K.U.Leuven, Department of Computer Science, Belgium
Mario Cannataro Informatics and Biomedical Engineering,
Department of Experimental and Clinical Medicine,
Faculty of Medicine, University "Magna Graecia" of Catanzaro, Italy
Vítor Santos Costa COPPE/Sistemas, UFRJ, Brazil and LIACC, Univ. Porto, Portugal
Inês Dutra COPPE/Sistemas, UFRJ, Brazil
João Gama LIACC and FEP, Universidade do Porto, Portugal
Alípio Jorge LIACC and FEP, Universidade do Porto, Portugal
Fernando Silva LIACC and FCUP, Universidade do Porto, Portugal

Sponsors

The workshop is sponsored by KDnet Logo left