Workshop on Applying case-based reasoning to time series prediction
In conjunction with ICCBR 2003, Trondheim, 24 June 2003
(Submission deadline 21 March 2003)
http://www-lipn.univ-paris13.fr/~kanawati/CbrSeqWS

I- Workshop objectives

Time series prediction is an important application field of artificial
intelligence (A.I.) techniques. A variety of real life applications require
predicting the evolution of times series. Application covers various fields
such as financial markets, medical diagnosis, meteorological forecasting,
forest fire forecasting, recommender systemsand web navigation advisors.
Different A.I. techniques have been applied to predicting the evolution of
time series. Examples of such techniques are artificial neural networks, fuzzy
logic, evolutionary algorithms and soft computing (the integration of the
above-cited techniques). Recently, a number of researchers from the Case-based
reasoning (CBR) community have addressed the problem of applying the CBR
problem solving methodology for predicting sequence or time series evolutions.
The different CBR systems that have tackled this problem each focus on a
particular issue. This workshop intends to bring together researches and
system developers to discuss and share their ideas about how to adapt the CBR
methodology for the specific task of time series prediction. The workshop is
expected to produce a first state of the art about using CBR for time series
treatment.

Topics relevant to the workshop include but are not restricted to:

- Comparing CBR with other A.I techniques. In which fields the use of CBR
methodology is likely to enhance the results of other A.I or statistical
techniques? Could we define benchmarks and specific evaluation criteria for
evaluating CBR based time series prediction?

- Case edition and representation. How to extract cases from raw sequences?
Does a case represent a whole sequence, or a part of sequence? Which
representation scheme to use in which application: interval representation or
point representation?

- Similarity measures. How to efficiently take into consideration the
sequential nature of cases when computing similarities?

- CBR cycle adaptation. How to exploit the sequential nature of cases in order
to enhance the different CBR phases including maintenance. The fact that the
Target case is also extracted from a sequence should allow to reuse reasoning
results obtained for past target cases extracted from the same sequence. What
techniques and heuristics could be applied to achieve this?

- Hybrid systems. What other techniques could be used in order to enhance CBR
prediction systems (artificial neural networks, fuzzy logic, etc.)? What type
of hybrid systems is more suitable for which CBR phase?

II-Paper submission

Participation in the workshop will be by invitation only. People interested in
participating in the workshop should submit either a full paper (10 pages) or
a position paper (2 pages). submissions should be formatted according to
Springer LNCS format, which is the format required for the final camera ready
copy. Author's instructions along with LaTeX and Word macro files are
available at <http://www.springer.de/comp/lncs/authors.html>. Submissions as
well as the names of potential participants must be emailed to :
[EMAIL PROTECTED] 21 April 2003. Accepted papers will be
included in the workshop notes and will be available on the workshop web site.
In order to have a highly interactive workshop, participation will be limited
to 25 people.

III-Workshop organisation

The workshop organisers will be charged to produce a pre-workshop report that
sums up the different accepted papers. The pre-workshop report will be
available to all participants three weeks before the workshop day. At most two
different relevant topics will be selected from the set of accepted
contributions. These issues will be advertised before the workshop. Each
participant will be asked to give her/his positions on each issue. The
workshop will be organised into three sessions: the first session is dedicated
for presenting the pre-workshop report. Both left sessions are dedicated to
discuss selected discussion topics. Considered issues will depend on the
topics addressed by the different accepted contributions. A post workshop
report will sum up the different contributions during the workshop.

IV-Important dates

21/03/2003      Deadline for paper submission
21/04/2003      Authors notification
05/05/2003      Camera ready
15/05/2003      Participation request deadline
1/06/2003       Pre-workshop report delivery
24/06/2003      Workshop day

V-Organisation committee

Rushed Kanawati, LIPN - University of Paris 13 [EMAIL PROTECTED]
Maria Malek, LAPI- EISTI [EMAIL PROTECTED]
Sylvie Salotti, LIPN - University of Paris 13 [EMAIL PROTECTED]

VI- Program committee

Juan Corchado, Universidad de Salamanca [EMAIL PROTECTED]
Martha Dorum Jare, Borak Desarollo S.L [EMAIL PROTECTED]
Rushed Kanawati, University of Paris 13 [EMAIL PROTECTED]
Brian Lees, University of Paisley [EMAIL PROTECTED]
Maria Malek, LAPI - EISTI [EMAIL PROTECTED]
Sankar K. Pal, Indian Statistical Institute [EMAIL PROTECTED]
Sylvie Salotti, University of Paris 13 [EMAIL PROTECTED]



-------------------------------------------------
Message envoye par IMP: http://horde.org/imp/
LIPN - CNRS UMR 7030 - http://www-lipn.univ-paris13.fr

-------------------
MLnet community list
http://www.mlnet.org/mlnet2/services/mlnet-community.html

Reply via email to