(text)
Buczak, Anna, ANNIE '98 - ARTIFICIAL NEURAL NETWORKS IN ENGINEERING,
November 1 - 4, 1998, St. Louis, Missouri http://www.umr.edu/~annie
--
latest news, publications, tools, meetings, and other relevant items
in the Data Mining and Knowledge Discovery field.
KDNuggets is currently reaching over 4700 readers in 60+ countries
2-3 times a month.
Submissions relevant to data mining and knowledge discovery are welcome
and should be emailed to gps
in ASCII or HTML format.
A submission should have a subject line which clearly describes
what is it about. Please keep calls for papers and meeting announcements
short (50 lines of up to 80 characters each),
and provide a web site for details. Submissions may be edited for size.
Commercial submissions are subject to a charge.
See kdnuggets.com/submissions.html for full guidelines.
Back issues of KD Nuggets, a catalog of data mining tools
('Siftware'), pointers to data mining companies, relevant websites,
meetings, and more is available at KDNuggets Directory at http://www.kdnuggets.com/
********************* Official disclaimer ***************************
All opinions expressed herein are those of the contributors and not
necessarily of their respective employers (or of KD Nuggets)
*********************************************************************
~~~~~~~~~~~~ Quotable Quote ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
A file that big?
It might be very useful.
But now it is gone.
from Salon http://www.salon1999.com/
Haiku Error Messages Previous1NextTop
Date: Tue, 17 Mar 1998 10:06:35
From: Gregory Piatetsky-Shapiro gps@kstream.com
Subject: KDD-98 Submissions
According to latest counts from AAAI, we have received about 240 papers for
KDD-98 (about 50% increase compared to last year !).
KDD-98 will also have a full range of tutorials,
panels, workshops, invited talks, and exhibits, and more, so all
indications are for a very exciting and interesting conference in August in
New York City.
Mining Your Own Business -- Vendors seek to
ease deployment as more companies look to
data mining to turn data into profits
March 17, 1998
Information Week cover story, March 16, entitled 'Mining Your Own Business',
discusses recent developments in the data mining market.
It talks about applications at
ITT Sheraton, the National Association of Securities Dealers, Safeco, and Wells Fargo
that range from identifying
which hotel guests might want a cigar in their rooms to detecting stock market fraud.
The story also describes plans for leading data mining software developers,
including DataMind, IBM, NeoVista, SAS Institute, and Silicon
Graphics, to introduce new packages that are increasingly being tuned
for specific applications such as fraud detection or customer-relationship
management and for vertical industries.
Also, Oracle is announcing partnership with
Angoss, DataMind, Datasage, Information Discovery, SPSS, SRA International, and
Thinking Machines to integrate their products into Oracle 8.1.
Microsoft electronic-commerce system, Site Server 3.0 Commerce edition,
due next quarter, will include Intelligent Cross Sell, a
data mining algorithm that will analyze the activity of shoppers on a Web site
and automatically adapt the site to that user's preferences.
'My vision is that this kind of information can be used to completely optimize a
Web store, to reorganize the appearance of a store to serve you better,'
says Usama Fayyad, a senior researcher with Microsoft Research,
which developed the feature. A Microsoft Research project called Socrates is studying
the potential use of data mining in very large databases running on Microsoft platforms.
the interview with Gregory Piatetsky-Shapiro, as published in
Hong Kong I.T. Times, March 11, 1998 -- (beware -- it is in Chinese),
and here is the full text of the interview
Many thanks to Stephen Koo for making the text available.
Previous4NextTop
Date: Wed, 11 Mar 1998 22:50:13 -0500
From: Maria Zemankova mzemanko@nsf.gov
Subject: NSF/CISE KDI/New Challenges to Computation information
Title: New Challenges to Computation (NCC) -- Dear Colleague
Letter
Type: Program Announcements & Information
Subtype: Computer/Information Sciences
It refers to the Computer and Information Science and Engineering (CISE)
focus of the NCC component of the Knowledge and Distributed Intelligence
initiative http://www.ehr.nsf.gov/kdi/.
Letters of intent due: April 1, 1998
Full proposals due: May 8, 1998 Previous5NextTop
Date: Thu, 19 Mar 1998 09:21:36 -0300
From: alex@dainf.cefetpr.br
(Alex Alves Freitas)
Subject: new book on parallel data mining
KLUWER ACADEMIC PUBLISHERS IS PROUD TO ANNOUNCE THE PUBLICATION OF...
MINING VERY LARGE DATABASES WITH PARALLEL PROCESSING
by
Alex A. Freitas, CEFET-PR, Dep. de Informatica, BRAZIL
Simon H. Lavington, University of Essex, UK
MINING VERY LARGE DATABASES WITH PARALLEL PROCESSING addresses the
problem of large-scale data mining. It is an interdisciplinary text,
describing advances in the integration of three computer science areas:
- 'intelligent' (machine learning-based) data mining techniques;
- relational databases;
- and parallel processing.
The basic idea is to use concepts and techniques of the latter two areas -
particularly parallel processing - to speed up and scale up data mining
algorithms. Included are:
- a comprehensive review of intelligent data mining techniques, such as
rule induction, instance-based learning, neural networks and genetic
algorithms
- a comprehensive review of parallel processing and parallel databases
- an overview of commercially-available, state-of-the-art tools
- the application of parallel processing to data mining
- cost-effective solutions for realistic data volume
- a discussion of two parallel computational environments
This volume will be a valuable source to industry data miners and
practitioners in applying intelligent data mining techniques to large
amounts of data. In addition, this book will be useful to academic
researchers and postgraduate students interested in advanced,
intelligent database applications and artificial intelligence
researchers interested in industrial real-world applications of
machine learning.
TABLE OF CONTENTS
Preface. Acknowledgments. Introduction. Part I: Knowledge
Discovery and Data Mining. 1. Knowledge Discovery Tasks. 2.
Knowledge Discovery Paradigms. 3. The Knowledge Discovery Process.
4. Data Mining. 5. Data Mining Tools. Part II: Parallel Database
Systems. 6. Basic Concepts on Parallel Processing. 7. Data
Parallelism, Control Parallelism and Related Issues. 8. Parallel
Database Servers. Part III: Parallel Data Mining. 9. Approaches to
Speed Up Data Mining. 10. Parallel Data Mining Without DBMS
Facilities. 11. Parallel Data Mining with Database Facilities. 12.
Summary and Some Open Problems. References. Index.
1998 224 pp. ISBN 0-7923-8048-7 $105.00
FOR MORE INFORMATION ABOUT THIS PUBLICATION, PLEASE VISIT OUR
On-line Catalogue at: http://www.wkap.nl
New Version Improves Analytical, Visual and Performance Capabilities
for Decision Support Solutions
MOUNTAIN VIEW, CA (March 11, 1998) -- Silicon Graphics, Inc. (NYSE:
SGI) today announced MineSet(TM) 2.5, the newest version of its fully
integrated, comprehensive suite of easy-to-use analytical and visual
data mining tools. MineSet 2.5 software tools revolutionize customers'
decision support process by offering parallelized data mining
algorithms for faster performance as well as new analytical tools,
such as regression, clustering, and decision tables for more intuitive
comprehension of data.
...
'MineSet provides executives with the analytical and visual insight
necessary to make critical decisions about their businesses,' said
Aaron Zornes, vice president of Application Delivery Strategies
Services at META Group. 'It is evident that with the new features in
MineSet 2.5, Silicon Graphics has created one of the best analytical
and visual data mining solutions available today.'
Customers that buy MineSet 2.01 now, before MineSet 2.5 releases are
eligible to receive a FREE upgrade to version 2.5 and avoid the 10% price
increase.
Updates from the MineSet Team
-----------------------------
1. Silicon Graphics customer education offers hands-on MineSet training
in small classes with a Silicon Graphics O2 workstation per person. See http://mineset.sgi.com/training
for details. The March 24-26
course is full. Open slots still available for April 29-May 1
and June 2-4, 1998, where MineSet 2.5 will be taught.
3. Silicon Graphics just released a report on MineSet supporting the Discovery
Research process in in Bioinformatics and Cheminformatics. See http://www.sgi.com/chembio/resources/mineset/
The report includes visualizing Genomic databases, Gel
Electrophoresis, Splice Junction prediction, exploring the Tripos
database of 180 billion compounds, and Structure Activity Relationships.
4. A new updated 'Guided tour of MineSet 2.0 using a Churn in
Telecommunications as an example' is available at http://mineset.sgi.com/contact.html
The talk is 'live' and requires a Silicon Graphics machine.
AUTOSPEC THEMATICS: conceptual media filters. Design of pocket
vocabularies for Internet, intranets, agents, research,
Special Interest Groups (SIGs). Relevance in context. Stan Rice http://www.cruzio.com/~autospec/.
Email: autospec@cruzio.com
>>FOR YOUR INTEREST IN CONTEXTUAL & CONCEPTUAL MEDIA FILTERING:
'Thematic vocabularies' are keys to conceptual media access, to
markets, etc. To pin-point Internet pages instantly, like those
below, e.g. for fuzzy boolean search? HOTBOT: FUZzy BOOlean (sic)
------------------------------------------------------------
SOME LINKS TO EASY THEMATIC CONCEPTUAL FILTERING
------------------------------------------------------------
For a home page on Thematic principles, examples, and links: http://www.cruzio.com/~autospec/
For bio information on yours truly, Whosis: http://www.cruzio.com/~autospec/srbio.htm
----------------------------------------------------------
If you care to share reactions, they are always gratefully
received. Best wishes, Stan Rice autospec@cruzio.com
ISoft, European Leader in Data Mining for business users is proud to announce
ALICE d'ISoft version 5.0
------------------------------------------------------------------
ISoft's high profile Data Mining product explores databases through
interactive decision trees and creates queries, reports, charts and rules
for predictive models. Designed and priced for the non-technical user,
ALICE d'ISoft gives business users access to the knowledge hidden in their
databases, discovering the trends and relationships in their data and
making predictions using that information.
ALICE d'ISoft v5.0 new features are:
ALL-IN-ONE DATA MINING WINDOW displays interactive decision trees,
interactive information sheet and On-Line Graphics. The decision trees
display global information while the information sheet and On-Line Graphics
panel focus on one selected node. The three are dynamically linked together
to provide consistent information at any time.
ON-LINE GRAPHICS enable users to catch at a glance and for any node the
structure of the node and the repartition of each field.
DIRECT DATA MANAGEMENT allow you to refresh your data at the click of the
mouse. Your data can be exported in text format, as a SQL request, or in
the clipboard. Three SQL formats are available: Access SQL, ANSI SQL and
Plain SQL.
TOOL-TIP INFORMATION BOXES appear on every significant object of the window
(node, OLG graph, etc.)
--------------------------------------------------------------------
For more information, visit http://www.alice.fr
or contact Raphaelle Thomas
at rthomas@isoft.fr.
-------------------------------------------------------------
Mme Raphaelle THOMAS ISoft
International Devt Manager Chemin du Moulon
Tel: +33 (0)1 69 35 37 37 91190 Gif sur Yvette
Fax: +33 (0)1 69 35 37 39 France
Web: http://www.alice.fr Previous9NextTop
Date: Thu, 19 Mar 1998 14:05:16 +0200 (EET)
From: Henry Tirri tirri@cs.Helsinki.FI
Subject: BAYDA 1.0 - free software for Bayesian classification
Web: http://www.cs.Helsinki.FI/research/cosco
BAYDA 1.0
Complex Systems Computation Group (CoSCo) announces the release of
BAYDA 1.0 software for Bayesian classification and feature selection
for discrete data
BAYDA (Bayesian Discriminant Analysis) is a Java software package for
flexible data analysis in classification tasks for discrete data. For
predicting the class memberships, BAYDA performs fully Bayesian
predictive inference based on a Naive Bayes model with the marginal
likelihood predictive distribution. As demonstrated in several
studies, using model parameter averaging improves classification
performance substantially, especially with small samples.
It is well-known that the Naive Bayes classifier performs well in
terms of prediction accuracy, when compared to approaches using more
complex models. However, the model makes strong independence
assumptions that are frequently violated in practice. For this reason,
the BAYDA software also provides a feature selection scheme which can
be used for analyzing the problem domain, and for improving the
prediction accuracy of the models constructed by BAYDA. The feature
selection can be done either manually or automatically. In manual
selection the user has an opportunity to use BAYDA for evaluating
different feature subsets by leave-one-out crossvalidation scheme. In
the automatic feature selection case the program selects the relevant
features by using the Bayesian supervised marginal likelihood
criterion.
The current version features of BAYDA include
- missing data handling
- an external leave-one-out crossvalidated estimate of the classifier
performance in graphical format
- 'intelligent document' style graphical interface
- forward selection/backward elimination feature subset selection
- free format data files (such as tab-delimited format of SPSS)
BAYDA is available free of charge for research and teaching purposes from
under section 'Software', and it has currently been tested on
Windows'95/NT, SunOS and Linux platforms. However, being implemented
in 100% Java, it should be executable on all platforms supporting Java
Runtime Environment 1.1.3 or later. Previous10NextTop
Subject: AVAILABLE: ROC Convex Hull program for comparing classifiers
From: Tom Fawcett fawcett@Basit.COM
Date: 19 Mar 1998 14:33:02 -0500
----------------------------------------------------------------
NOW AVAILABLE: ROC Convex Hull program for comparing classifiers
----------------------------------------------------------------
In our data mining/machine learning work we often face domains in
which class distributions are greatly skewed and/or classification
error costs are unequal. In these situations, the evaluation of
classifiers is very difficult because classification accuracy, the
metric by which most evaluation is currently done, is completely
inadequate. To make things worse, class distributions in these
domains often drift over time, and error costs may be known only
approximately.
We've developed a robust framework for evaluating learned classifiers,
based on ROC analysis, which enables us to analyze and visualize
classification performance separately from assumptions about class
distributions and error costs.
The method computes the ROC convex hull and allows us to:
- analyze classifier performance over a broad range of performance
conditions (error costs and target class distributions),
- determine the range of conditions under which a given classifier
will be best, and
- determine easily the best available classifier(s) for any particular
conditions.
We now use this method extensively in our applied work as well as in
our research on classification, and other researchers have begun using
it as well. We've decided to place the program under the Gnu Public
License (GPL) and make it available to the ML and Data Mining
communities. The program and several papers on the technique are
available from:
'...I have finally got the experiments rolling for which I wanted to
use your ROC convex hull method -- I had a classic 'victory' with it
last night, a very clear picture emerging which probably no other
method of analysis would have uncovered, certainly not as clearly.'
-- Rob Holte
Previous11NextTop
Date: Wed, 11 Mar 1998 12:35:24 +0200 (IST)
From: 'Prof. Martin GOLUMBIC' golumbic@macs.biu.ac.il
Subject: Call for papers Workshop on KDD Bar-Ilan University May 20-21, 1998
C A L L F O R P A P E R S A N D P R E S E N T A T I O N S
Bar-Ilan Workshop on KDD -- Knowledge Discovery in Databases
May 20-21, 1998
Bar-Ilan University, Ramat-Gan, Israel
CALL FOR PAPERS/PRESENTATIONS
The Bar-Ilan Research Institute for Computer Science will sponsor a
workshop on Knowledge Discovery to be held May 20-21, 1998 at the
university. Submissions of short papers or presentations from academia
and industry are solicited in this Call.
Invited hour speakers (to date):
Haym Hirsh (Rutgers Univ.)
Oren Etzioni (Univ. Washington)
Ronen Feldman (Bar-Ilan Univ.)
Simon Kasif (Univ. of Illinios, Chicago)
Other participants to be announced.
Knowledge discovery from data is a broad discipline that integrates methods
from machine learning, statistics, databases, rule-based systems, and
other areas. It includes algorithms for data selection, pattern
discovery, clustering, managing uncertainty, and trend analysis.
Submissions are invited for research papers and presentations.
Topics to be covered include but are not limited to the following:
Text Mining
Pattern Matching for KDD
Rule Extraction
Algorithm Complexity and Lower Bounds
Incremental Discovery Methods
A short 1-3 page extended
abstract should be sent to Prof. Martin Golumbic, (golumbic@cs.biu.ac.il)
no later than April 20, 1998. Decisions for acceptance will be ongoing
and usually within 2 weeks of the submission. An on-line proceedings
of extended abstracts will be made available shortly before the workshop.
Date: July 8-10, 1998
Place: Katholieke Universiteit Leuven, Belgium
Organized at the Department of Electrical Engineering (ESAT-SISTA) and the
Interdisciplinary Center for Neural Networks (ICNN) in the framework of the
project KIT and the Belgian Interuniversity Attraction Pole IUAP P4/02.
In cooperation with the IEEE Circuits and Systems Society.
* GENERAL SCOPE
The rapid growth of the field of neural networks, fuzzy systems
and wavelets is offering a variety of new techniques for modeling
of nonlinear systems in the broad sense. These topics have been
investigated from differents points of view including statistics,
identification and control theory, approximation theory, signal
processing, nonlinear dynamics, information theory, physics and
optimization theory among others. The aim of this workshop is to serve
as an interdisciplinary forum for bringing together specialists in these
research disciplines. Issues related to the fundamental theory as well
as real-life applications will be addressed at the workshop.
* TIME-SERIES PREDICTION COMPETITION
Within the framework of this workshop a time-series prediction
competition will be held. The results of the competition will be
announced during the workshop, where the winner will be awarded.
Participants in the competition are asked to submit their predicted
data together with a short description and references of the
methods used. In order to stimulate wide participation in the
competition, attendance of the workshop is not mandatory but
is of course encouraged. All information about this contest is available
at http://www.esat.kuleuven.ac.be/sista/workshop/
.
* IMPORTANT DATES
Deadline paper submission: April 2, 1998
Notification of acceptance: May 4, 1998
Workshop: July 8-10, 1998
<>
Previous13NextTop
Subject: ECML'98 - Call for Participation
Date: Mon, 16 Mar 1998 17:37:14 +0100
From: Conf ECML98 Conf.ECML98@lri.fr
Call for Participation
TENTH EUROPEAN CONFERENCE ON MACHINE LEARNING (ECML'98)
Chemnitz, Germany, April 21-24 1998
-------------------------------------------------------------------------
Up-to-date information on the conference can be found at
The 10th European Conference on Machine Learning (ECML'98) will be
held in Chemnitz (ex- Karl Marx Stadt, near Dresden and Berlin),
Germany, from April, 21st to 24th 1998.
PROGRAM
The scientific program (April 21 - 23) will include invited talks,
presentations of accepted papers, poster and demonstration
sessions. The call for poster and demonstration is open until 25 March
(see http://www.lri.fr/~ecml98/poster-demo.html
for more details).
Saturday, April 24, will be devoted to workshops. The conference
proceedings will be published by Springer Verlag, Berlin, as part of
the 'Lecture Notes in AI (LNAI)' series. Detailed information
regarding the scientific program and the workshops can be found on
the ECML'98 web page http://www.tu-chemnitz.de/informatik/ecml98/
Topics to be addressed in conference presentations include:
Applications of ML Inductive Logic Programming
Bayesian Networks Relational Learning
Feature Selection Instance-Based Learning
Decision Trees Clustering
Support Vector Learning Genetic Algorithms
Multiple Models for Classification Reinforcement Learning
Neural Networks
<>
For full information and registration see the web site.
The Second International Conference on Practical Aspects
of Knowledge Management (PAKM98)
29-30 October, 1998
Basel, Switzerland
Supported by
SGAICO (Swiss Group for Artificial Intelligence and Cognitive Science)
and the
Special Interest Group 'Knowledge Engineering' of the German Computer Society
Aims and scope of the conference
--------------------------------
It is widely acknowledged that knowledge is one of the most important assets
of organizations. Especially companies in industrialised countries with high
wages can only compete on the global market when offering products that are
based on advanced technology or when trading the technology itself, thus
having an advantage over companies in countries with low salaries. These
companies depend on highly educated and skilled employees as well as on
short innovation cycles, high flexibility and creativity. One of the
prerequisites to achieving this is a systematic management of the key
success factor 'knowledge'.
Knowledge Management is primarily an issue of enterprise organization and
enterprise management but there are many central and important issues which
can be supported or even enabled by state-of-the-art information systems.
Consequently, approaches to Knowledge Management need to be rooted in
business and organization science as well as in computer science. However,
conferences and workshops on Knowledge Management typically either cover
approaches from the first or the second area only. Although such events are
certainly worthwhile we feel that bringing together people from both areas
and giving them a forum for exchanging ideas will lead to Knowledge
Management solutions that are much more useful and effective.
The PAKM Conference is dedicated to that quite challenging aim. It will
bring together people from both areas, namely
* people who have an organizational perspective on Knowledge Management,
e.g. have practical experience in introducing Knowledge Management in
organisations, or are concerned with more theoretical approaches to
managing the resource 'knowledge'
* people with an information technology point of view on Knowledge
Management who, e.g., have developed tools for Knowledge Management, or
are investigating on a more theoretical level technological frameworks
for Knowledge Management
Call for Papers: UNCERTAINTY 99
Seventh International Workshop on Artificial Intelligence and Statistics
January 3-6, 1999,
Ft. Lauderdale, Florida http://uncertainty99.microsoft.com/
This is the seventh in a series of workshops which has brought
together researchers in Artificial Intelligence (AI) and in Statistics
to discuss problems of mutual interest. The exchange has broadened
research in both fields and has strongly encouraged interdisciplinary
work. Papers on all aspects of the interface between AI & Statistics
are encouraged.
To encourage interaction and a broad exchange of ideas, the
presentations will be limited to about 20 discussion papers in single
session meetings over three days (Jan. 4-6). Focused poster sessions
will provide the means for presenting and discussing the remaining
research papers. Papers for poster sessions will be treated equally
with papers for presentation in publications. Attendance at the
workshop will not be limited.
The three days of research presentations will be preceded by a
day of tutorials (Jan. 3). These are intended to expose researchers in
each field to the methodology used in the other field. The tutorial
speakers will include
Chris Bishop, Cambridge,
Latent variables and neural networks.
Sue Dumais, Seattle,
Information access and retrieval.
and the keynote speaker is
David Spiegelhalter, Cambridge, on
Bayesian statistical analysis.
SMART ENGINEERING SYSTEM DESIGN
Neural Networks, Fuzzy Logic,
Evolutionary Programming, Data
Mining and Rough Sets
Organizer: UNIVERSITY OF MISSOURI-ROLLA
In Cooperation with IEEE NEURAL NETWORKS COUNCIL
November 1 - 4, 1998, Marriott Pavilion Hotel, St. Louis, Missouri
SMART ENGINEERING SYSTEM DESIGN: NEURAL NETWORKS, FUZZY LOGIC,
EVOLUTIONARY PROGRAMMING, DATA MINING, AND ROUGH SETS
ANNIE '98 will be held on November 1-4, 1998, at Marriott's
Pavilion Hotel in downtown St. Louis, Missouri, USA. This will
be the eighth international gathering of researchers interested
in Smart Engineering System Design using neural networks,
fuzzy logic, evolutionary programming, data mining, and rough
sets. The previous conferences each drew approximately 150
papers from twenty countries. The proceedings of all
conferences were published by ASME Press as hardbound
books in seven volumes. The last volume, edited by Dagli, et.
al., was titled 'Smart Engineering Systems: Neural Networks,
Fuzzy Logic, Data Mining and Evolutionary Programming'.
ANNIE' 98 will cover the theory of Smart Engineering System
Design techniques, namely; neural networks, fuzzy logic,
evolutionary programming, data mining, and rough sets.
Presentations dealing with applications of these technologies
are encouraged in the areas of: manufacturing engineering,
biology and medicine, pattern recognition, image processing,
process monitoring, control, recent theoretical developments in
neural networks, fuzzy logic, data mining, rough sets,
evolutionary programming, fractals, chaos, and wavelets that
can impact smart engineering system design.
CALL FOR CONTRIBUTED PAPERS
The organizing committee invites all persons interested in
Smart Engineering System Design using neural networks, fuzzy
logic, evolutionary programming, data mining, and rough sets
to submit papers for presentation at the conference. All papers
accepted for presentation will be published in the conference
proceedings.