News: *
GPS, KDD-97 Best Paper Awards: The winners are ... *
John Thompson, Terabyte Challenge News from Magnify Inc. Publications: *
Se June Hong, Special issue of FGCS on Data Mining *
Torulf Mollestad, PhD on Rough Sets Data Mining
*
John R. Koza, Genetic Programming Conference (GP-98),
July 22 - 25, 1998, Madison, Wisconsin, www.genetic-programming.org
--
Data Mining and Knowledge Discovery community, focusing on the
latest research and applications.
Submissions are most welcome and should be emailed, with a
DESCRIPTIVE subject line (and a URL) to gps.
Please keep CFP and meetings announcements short and provide
a URL for details.
KD Nuggets frequency is 3-4 times a month.
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 Knowledge Discovery Mine site
at
********************* Official disclaimer ***************************
All opinions expressed herein are those of the contributors and not
necessarily of their respective employers (or of KD Nuggets)
*********************************************************************
~~~~~~~~~~~~ Quotable Quote ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
They are ill discoverers that think there is no land when
can see nothing but sea.
Francis Bacon Previous1NextTop
Date: Thu, 31 Jul 1997 09:32:16 -0400
Subject: KDD-97 Best Paper Awards: The winners are ...
From: Gregory Piatetsky-Shapiro (gps@kstream.com)
On behalf on KDD-97 Awards Committee I am happy to
announce the winners of KDD-97 Best Paper Awards, sponsored by
Knowledge Stream Partners.
In the Fundamental Research category,
the winners are
Foster Provost and Tom Fawcett, NYNEX Science and Technology Center,
for the paper
Analysis and Visualization of Classifier Performance:
Comparison under Imprecise Class and Cost Distributions
(the runner-up paper was
A Probabilistic Approach to Fast Pattern
Matching in Time Series Databases
Eamonn Keogh and Padhraic Smyth, University of California, Irvine)
In the Applied Research category,
the winners are
Padhraic Smyth, University of California, Irvine;
Michael Ghil and Kayo Ide, University of California, Los Angeles;
Joe Roden, Jet Propulsion Laboratory, California Institute of Technology;
Andrew Fraser, Portland State University,
Detecting Atmospheric Regimes using Cross-Validated Clustering
(the runner-up paper was
JAM: Java Agents for Meta-Learning over Distributed Databases.
by Sal Stolfo , A. Prodromidis, and P. Chan)
The awards will be presented at KDD-97.
Gregory Piatetsky-Shapiro, KDD-97 Best Paper Awards Chair
Knowledge Stream Partners
KDD-97 Best Paper Awards Committee
Tej Anand
Ted Senator
Brij Masand
Gregory Piatetsky-Shapiro
Graham Wills
Wojtek Ziarko
Charles Elkan
Se June Hong
David Jensen
Full KDD-97 information, including the program is at
Previous2NextTop
From: John K. Thompson (jkt@magnify.com)
Subject: Terabyte Challenge News from Magnify Inc.
Date: Wednesday, July 30, 1997 8:29 AM
Contact: John K. Thompson
(312) 214-1420, (312) 214-1429 FAX
Magnify, Inc.
100 South Wacker Drive
Suite 1130
Chicago, IL 60606
---
Model Interchange Format by Magnify, Inc.
Magnify's test of proposed Model Interchange Format proves viability of
open standards for sharing of predictive models.
Chicago, Ill. -July 24, 1997-Magnify, Inc. announced today the
successful completion of the first round of tests to validate an open,
standards based approach to storing, transporting, and sharing
predictive models.
The tests were conducted in conjunction with the Terabyte Challenge.
The Terabyte Challenge is a year long technology demonstration of high
performance data mining and data intensive computing on massive data
sets involving university and commercial scientists.
Magnify has written the Model Interchange Format (MIF) specification
with the intention of publishing the specification for use by all
vendors who are producing predictive models. The MIF file structure is
a pioneering concept built by Magnify to allow data mining models to be
built and stored in an open format that provides for portability and
flexibility.
'The proprietary nature of the available data mining systems is
inhibiting the growth of the data mining industry as a whole,' said
Robert Grossman, Ph.D., President of Magnify, Inc. 'Clients are
beginning to look to the industry innovators to provide mechanisms
whereby they can exchange models between business units, locations, and
affiliated companies, as well as facilitating the use of multiple
products from various mining vendors. Magnify will meet this need with
the MIF specification.'
As part of the distributed data-mining test within the Terabyte
Challenge, Magnify engineers verified the operability of the MIF
specification. 'Models were built, tested, and stored in MIF format. We
sent the models to the remote system and after the MIF's were received,
and stored, the models worked just as if they had never been moved,'
said Ivan Pulleyn, Senior Member of the Technical Staff at Magnify, Inc.
After further tests and refinement, the MIF specification will be
presented to vendors and other interested parties for review and
consideration. The industry will grow at increased rates once clients
can leverage their investments in existing modeling techniques and
technologies by augmenting those investments with the most appropriate
innovations as those innovations are made available by the leaders in
the data mining industry.
---
Next Generation Internet provides conduit for Distributed Data Mining.
Chicago, Ill. -July 24, 1997-Magnify, Inc. in conjunction with two
leading academic institutions have completed tests over a wide area
network that prove the viability of performing data mining in a
distributed mode.
The tests of distributed data mining and improved video conferencing
were performed within the framework of the Terabyte Challenge. The
Terabyte Challenge is a year long technology demonstration of high
performance data mining and data intensive computing on massive data
sets involving university and commercial scientists.
'Although it is becoming common today to have data warehouses that are
over a terabyte in size, it is still an open issue of how to most
efficiently perform the critical tasks of data management, analysis and
data mining on those very large data sets,' said Robert Grossman, Ph.D.,
President of Magnify, Inc. 'Working with the resources and personnel of
the Terabyte Challenge allows us to push the capacity limits to the
petabyte range and begin to solve data management and data mining issues
that will be problems confronting our clients in the commercial sector
in the coming year.'
While the majority of current efforts have focused on data management
and data mining in a centralized mode, the recent and dramatic growth of
the Web has directed leading edge researchers and commercial developers
to focus their attention on work required to locate, integrate, and make
effective use of the distributed information now available on wide area
networks. These efforts represents forefront of technological
innovation in the areas of data management, information retrieval, and
high performance distributed data mining.
---
About Magnify, Inc.
Founded in 1991, Magnify, Inc. provides innovative, scalable data mining
solutions for managing, mining, and analyzing large amounts of data.
Magnify, Inc. and Magnify Research, Inc. (Magnify's sister company which
develops products and integrates systems for the federal government) are
subsidiaries of Magnify Holdings Corporation. For more information
access Magnify's Web site at
Trademarks or registered trademarks of Magnify Holdings Corporation, or
its subsidiaries are: Magnify, PATTERN, PATTERN:Detect, PATTERN:Profit,
and the Magnify logo. All other brands and product names are trademarks
or registered trademarks of their respective companies.
About the Terabyte Challenge
The Terabyte Challenge is an evolving, open test-bed that can be used to
test new algorithms and software for high performance and wide area
data; management, mining and analysis. One of the major objectives of
the Terabyte Challenge is to understand the best technology for the next
step of managing, analyzing, and mining petabytes of data (1000
Terabytes). For more information access the Supercomputing web site at:
Previous3NextTop
From: 'Se June Hong (8-862-2265)' (HONG@watson.ibm.com)
To: 'gps'
(gps)
Subject: nugget item: Special issue of FGCS on Data Mining
Date: Wed, 16 Jul 1997 16:20:37 -0400
FGCS (Future Generation Computer Systems), volume 13, Number 2, Oct 1997
Special Issue on Data Mining
CONTENTS
S.J Hong, 'Guest editorial: Data Mining'
U. Fayyad and P. Stolorz, 'Data Mining and KDD: Promise and challenges'
J. Hosking, E. Pednault and M. Sudan, 'A Statistical Perspective on
Data mining'
P. Michaud, 'Clustering Techniques'
M. Zait and H. Messatfa, 'A Comparative Study of Clustering Methods'
R. Srikant and R. Agrawal, Mining Generalized Association Rules'
I. Kononenko and S.J. Hong, 'Attribute Selection for Modelling'
C. Apte and S. Weiss, 'Data Mining with Decision Trees and Decision
Rules
M.W. Craven and J.W. Shavlik, 'Using Neural Networks for Data Mining'
B. Dunkel, N. Soparkar, J. Szaro and R. Uthurusamy, 'Systems for KDD:
From Concepts to Practice'
Abstracts
Data Mining and KDD: Promise and Challenges
U. Fayyad and P. Stolorz
Databases are growing in size to a stage where traditional
techniques for analysis and visualization of the data are breaking
down. Data mining and Knowledge Discovery in Data bases (KDD) are
concerned with extracting models and patterns of interest from
large large databases. Data mining techniques have their origins
in methods from statistics, pattern recognition, databases, artificial
intelligence, high performance and parallel computing, and
visualization. In this article, we provide an overview of this growing
multi-disciplinary research area, outline the basic techniques, and
provide brief coverage of how they are used in some applications. We
discuss the role of high performance and parallel computing in data
mining problems, and we provide a brief overview of a few applications
in science data analysis. We conclude by listing challenges and
opportunities for future research.
A statistical perspective on data mining
J. H. M. Hosking, E. P. D. Pednault and M. Sudan
Data mining can be regarded as a collection of methods for drawing
inferences from data. The a aims of data mining, and some of its
methods, overlap with those of classical statistics. However, there are
some philosophical and methodological differences. We examine these
differences, and we describe three approaches to machine learning
that have developed largely independently: classical statistics,
Vapnik's statistical learning theory, and computational learning
theory. Comparing these approaches, we conclude that statisticians
and data miners can profit by studying each other's methods and using
a judiciously chosen combination of them.
Clustering Techniques
P. Michaud
Given a population of individuals described by a set of attribute
variables, clustering then into 'similar' groups has many applications.
The clustering problem, also known as unsupervised learning, is the
problem of partitioning a population into clusters (or classes). The
population is a set of n elements that can be clients, products, shops,
agencies etc., described by m attributes. These attributes can be
quantitative (salary), categorical (type of profession) or binary
(owner of a credit card). The goal is to construct a partition in
which elements of a cluster are 'similar' and elements of different
clusters are 'dissimilar' in terms of the m attributes. Here we
define the clustering problem and discuss the ideas behind some of
the major approaches, including a relatively new method, called
RDA/AREVOMS, that is based on the theory of voting.
A Comparative Study of Clustering Methods
M. Zait, H. Messatfa
In this paper we propose a methodology for comparing clustering methods
based on the quality of the result and the performance of the execution.
We applied it to several known clustering methods: FastClust, Autoclass
Relational Data Analysis, and Kohonen nets. The quality of a clustering
result depends on both the similarity measure used by the method and
its implementation. An important feature of our methodology is a
synthetic data generation program that allows producing datasets with
specific (or desired) patterns using a combination of parameters, such
as the number and the type of attributes, the number of records, etc.
We define a metric to measure the quality of a clustering method, i.e.
its ability to discover some or all of the 'hidden' patterns. The
performance study is based on the resource consumption, i.e., CPU
time and memory space.
Mining Generalized Association Rules
R. Srikaant, R. Agrawal
We introduce the problem of mining generalized association rules.
Given a large database of transactions, where each transaction
consists of a set of items, and a taxonomy (is-a hierarchy) on the
items, we find associations between items at any level of the taxonomy.
For example, given the taxonomy that says that jackets is-a outerwear
is-a clothes, we may infer a rule that 'people who buy jackets tend to
buy shoes', and 'people who buy clothes tend to buy shoes' do not hold.
An obvious solution to the problem is to add all ancestors of each item
in a transaction to the transaction, and then run any of the algorithms
for mining association rules on these 'extended transactions'. However,
this 'Basic' algorithm is not very fast; we present two algorithms,
Cumulate and EstMerge, which run 2 to 5 times faster than Basic ( and
more than 100 times faster on one real-life dataset). Finally, we
present a new interest-measure for rules which uses the information
in the taxonomy. Given a user-specified 'minimum-interest-level',
this measure prunes a large number of redundant rules; 40% to 60%
of all the rules were pruned on two real-life datasets.
Attribute Selection for Modelling
I, Kononenko, S.J. Hong
Modelling a target by other attributes in the data is perhaps the most
traditional data mining task. When there are many attributes in the
data, one needs to know which of the attribute(s) in machine learning.
We examine various important concepts and approaches that are used
for this purpose and contrast their strengths. Discretization of
numeric attributes is also discussed for its use is prevalent in
many modelling techniques.
Data Mining with Decision Trees and Decision Rules
C. Apte, S. Weiss
This paper describes the use of decision tree and rule induction in
data mining applications. Of methods for classification and
regression that has been developed in the fiends of pattern
recognition, statistics, and machine learning, these are of
particular interest for data mining since they utilize symbolic
and interpretable representations. Symbolic solutions can provide
a high degree of insight into the decision boundaries that exist
in the data, and the logic underlying them. This aspect makes these
predictive mining techniques particularly attractive in commercial
and industrial data mining applications. We present here a synopsis of
some major state-of-the-art tree and rule mining methodologies, as
well as some recent advances.
Using Neural Networks for Data Mining
M. Craven, J. Shavlik
Neural networks have been successfully applied in a wide range of
supervised and unsupervised learning applications. Neural-network
methods are not commonly used for data-mining tasks, however,
because they often produce incomprehensible models and require long
training times. In this article, we describe neural network
learning algorithms that are able to produce comprehensible
models, and that do not require excessive training times.
Specifically, we discuss two classes of approaches for data mining
with neural networks. The first type of approach, often called
rule extraction, involves symbolic models from trained neural
networks. The second approach is to directly learn simple, easy-to-
understand networks. We argue that, given the current state of the
art, neural-network methods deserve a place in the tool boxes of
data-mining specialists.
Systems for KKD: From Concepts to Practice
B. Dunke, N. Soparkar, J. Szaro, R. Uthurusamy
The considerable interest in knowledge discovery in databases (KDD)
has led to several techniques and tools for the automated
extraction of useful information from large data repositories. In
order to use these developments in practical settings, there is need
to consider the computing systems that would support the complete
KDD process. In this regard, we identify and discuss important
computing systems issues, and we compare some available research
and commercial efforts. Also, we suggest several enhancements to
the underlying database systems that may significantly benefit the
KDD process. We indicate why it is important that tools to handle
different aspects of the KDD effort need to be integrated. Finally,
we briefly describe our experience in implementing a prototype KDD
system for a large corporate environment.
Previous4NextTop
From: Torulf Mollestad (tm@neptun.computas.no)
Subject: PhD on Rough Sets Data Mining
Date: Wed, 23 Jul 1997 11:24:08 +0200
Dear Sirs,
I recently completed my PhD at the Norwegian University of Science and
Technology, Trondheim, Norway, on the
topic of Rough Sets data mining. The work, which was supervised by
Professor Dr. Jan Komorowski, is described
below and can be found at
A Rough Set Approach to Data Mining: Extracting a Logic of Default Rules
from Data
Torulf
Mollestad
Abstract
In this thesis, the problem of Data Mining (DM) is investigated, that
is, constructing decision rules from a set of
primitive input data. The starting point for the DM process is a table,
called an information system, which records
a number of objects according to specific attributes. One attribute is
singled out as the decision attribute, modelling
some expert's classification (``diagnosis'') of the object. The other
attributes are called conditional, and the problem
is to generate rules, defined over the latter set of attributes, that
reflect the expert classification.
The main contention of the present work is that there is a need to be
able to reason also in presence of inconsistencies,
and that more general, possibly unsafe rules should be made available
through the DM process. Such rules are
typically simpler in structure and allow the user to reason in absence
of information. Rough Set theory is used as the
underlying framework for mining of rules that reflect limited knowledge.
A framework is suggested for the automatic extraction
of propositional default rules that reflect normal intra-dependencies in
the data. The proposed algorithm introduces
indeterminacy by removing conditional attributes in a controlled manner.
The selection of attributes to be removed
is made from the factors in the discernibility function, thereby
removing information needed to discern classes in the
original information system. By this procedure, a number of different
default decision algorithms (sets of default rules) are
obtained, each of which classifies according to information over a
subset of the conditional attributes. Hence, when classifying
new cases, the default decision algorithm which is best suited to
handling the information at hand may be selected and applied.
The approach offers the possibility to direct an information gathering
process, through upward traversal in the lattice of information
systems. At each point, the attribute(s) may be selected that are
presumed to give the most information relative to the current
situation. In this light, a link is drawn to prioritised default
frameworks, arguing that an upward traversal in the lattice enables the
use of increasingly more specific rules. If the more specific rules are
in conflict with conclusions drawn on the basis of less
information then the latter conclusions are retracted.
A number of properties of the framework are investigated, with special
emphasis on methods for limiting an exponential search
space. Also, a framework is defined for making an exhaustive search for
functional dependencies in an information system; the
connection with the default rule extraction algorithm is intuitive.
A prototype implementation has been developed by project and diploma
students of the Knowledge Systems Group, and tests
have been run on several different data sets. The data was preprocessed
using two different systems for reasoning with Rough
Sets, namely Rses, developed at the Warsaw University, and Rosetta,
developed by members of the Knowledge Systems
Group, Norwegian University of Science and Technology. The results
suggest that default rules are good for classifying new
objects in situations of limited knowledge, and also that default rules
give a good view of the relative importance of the different
attributes. The knowledge is presented in an explicit way, in a manner
which is easily understandable to a human being.
Previous5NextTop
From: Huw Roberts (huw.roberts@bt-sys.bt.co.uk)
Subject: CFP: IST Special Issue on Knowledge Discovery and Data Mining
Date: Thu, 24 Jul 1997 17:10:33 +0100
Call For Papers:
Information and Software Technology (IST)
Special Issue on Knowledge Discovery and Data Mining.
We are soliciting papers of about eight journal pages each, though we
could have a few
up to 8,000 words in length if necessary.
Because you and other authors may find it impossible to submit by our
full paper deadline
of 15th February, 1998, we are issuing calls which could result in
substantially more submissions
than we have room for. So, if you can, please let us know by 1st
September 1997 if you intend to
submit. A refereeing process will take place on the basis of long
abstracts (of approximately
500 words in length), to be submitted by 1st October 1997. It will be
conducted by the two of us
calling on expertise from others where required. Authors will be
informed on 1st December 1997
of the outcome and the full paper will be required by 15th February 97.
Co-authored papers will be acceptable. Overviews, technical
contributions, reports on applications
and relatively speculative accounts of progress can all be considered.
The overall goal is to highlight
the challenges, excitement and progress to date of knowledge discovery
in databases, with the
objective of stimulating further research and applications.
Specific topics include, but are not restricted to:
Machine Learning, Statistical, etc, methods for use in Data Mining;
Algorithms for Reasoning Under Uncertainty about data;
Representation issues;
Technical Measures of Importance, Interestingness, etc, and their
evaluation;
Solutions to basic problems raised in developing DBMS functionality
and services to
support Data Mining better;
Incorporating domain knowledge and relevant process background
knowledge in Data Mining;
Problems caused by large volumes of data; scaleability of Data
Mining
Coping with updates to/dynamism in databases for Data Mining;
Data Mining for non-experts;
Data mining time-series data, text, multimedia data, etc.;
Tools and Applications of Data Mining.
IMPORTANT DATES:
ABSTRACTS DUE: 1st OCTOBER 1997
ACCEPTANCE NOTICES: 1st DECEMBER 1997
FULL PAPERS DUE: 15th FEBRUARY 1998
PUBLICATION: JUNE/JULY 1998
Inquiries and abstracts should be e-mailed (in ASCII format) to both of
us:-
GUEST EDITORS
Prof. D A Bell
Head of Information and Software Engineering
University of Ulster
UK
email: da.bell@ulst.ac.uk
H D Roberts
Data Mining Group
BT Laboratories
UK
email: huw.roberts@bt-sys.bt.co.uk
Data Analysis and Mining Software Tools
-----------------------------------------------------------------------------
Previous7NextTop
Date: Wed, 23 Jul 1997 14:54:44 +0400 (WSU DST)
From: Sergei Arseniev (megaputer@glas.apc.org)
Subject: New version of PolyAnalyst
Megaputer Intelligence announces a new version 3 of PolyAnalyst for
Microsoft Windows NT. FREE evaluation copy is available from our site:
www.megaputer.ru.
Sincerely,
Sergei Arseniev
Managing Director
Megaputer Intelligence, Ltd.
B. Tatarskaja 38,
113184 Moscow, Russia
Tel: 007 (095) 231-8079
Fax: 007 (095) 233-5371
E-mail: megaputer@glas.apc.org
internet: www.megaputer.ru
Previous8NextTop
From: Mark Embrechts (embrem@rpi.edu)
Date: July 28, 1997
Subject: Statistics Faculty position at RPI
URL:
DEPARTMENT OF DECISION SCIENCES AND ENGINEERING SYSTEMS
A tenure-track Assistant Professorship in statistical computing is available
commencing in September 1997:
Responsible for teaching courses at the graduate and undergraduate
levels and undertaking research in statistical computing and statistical
methodology. Possible areas of specialization include linear and non-linear
models, quality and reliability engineering, and data mining. Strong
interest in engineering applications is required.
The position requires outstanding research and teaching potential or record,
commitment to statistics in an academic engineering environment and excellent
communication skills.
Decision Sciences and Engineering Systems is one of ten departments within
the School of Engineering. With 21 regular faculty members and a number of
affiliated faculty from other departments, the department offers an
undergraduate degree in industrial and management engineering; master's
degrees in operations research and statistics, industrial and management
engineering, and manufacturing systems engineering; and a doctoral degree in
decision sciences and engineering systems. The department is heavily
involved in teaching within the core engineering program and is home to
Rensselaer's Statistical Consulting Center. The department is responsible
for some 150 undergraduate students, 100 masters students, and 40 doctoral
students.
Rensselaer Polytechnic Institute, founded in 1824, is the nation's oldest
technological university. With an undergraduate and graduate population of
6,000 students, Rensselaer is a co-educational, non-sectarian university and
multicultural university located in New York State's Capital District, a
thriving metropolitan area within 200 miles of New York, Boston and Montreal.
Interested individuals should send vitae and three letters of reference to:
Professor M. Raghavachari, Chair, Search Committee
Department of Decision Sciences and Engineering Systems
Rensselaer Polytechnic Institute
110 8th Street
Troy, NY 12180-3590
FAX: 518-276-8227
Rensselaer is an affirmative action/equal opportunity employer.
Women and minorities are encouraged to apply.
Previous9NextTop
From: Benedict Tanyi (tanyi@fecit.co.uk)
Subject: UK: FUJITSU European Centre For information technology (FECIT)
Date: Tue, 29 Jul 1997 16:34:00 +0100 (BST)
URL:
POSITION AT THE FUJITSU EUROPEAN CENTRE FOR
INFORMATION TECHNOLOGY (FECIT)
FECIT (a subsidiary of Fujitsu Ltd., Japan) is a
multidisciplinary research centre devoted to the development
of information technology on the latest high performance
parallel computers (visit
for more
information on FECIT's research activities).
Applications are invited from recent PhDs or suitably qualified
graduates for a Research Position in Data Mining/Warehousing
within the Financial Engineering Group at FECIT.
Applicants should be skilled in the latest Data Mining/Warehousing
technologies, in particular, Genetic Algorithms, Neural Networks,
statistics, etc. Experience in the use of high-performance parallel
architectures for Data Mining/Warehousing is highly desirable.
Excellent software development skills are required and a good general
mathematical background is essential.
It is also essential that the applicants have:
- the ability to do independent research;
- the ability and desire to work in teams of individuals
with diverse backgrounds;
- enthusiasm for working on applications;
- good communication skills.
Start Date: immediate or as soon as possible.
Salary : excellent remuneration package available.
Informal Enquiries can be directed to:
Dr. Koji Tajima Dr. Benedict Tanyi
E-mail: tajima@fecit.co.uk
E-mail: tanyi@fecit.co.uk
Tel: +44(0)181-606-4520 Tel: +44(0)181-606-4444 Ext. 2151
Qualified candidates should send their CVs to:
Mrs. Edna Davis
Fujitsu European Centre for Information Technology Ltd
2 Longwalk Road
Stockley Park, Uxbridge
Middlesex UB11 1AB
United Kingdom.
The Eleventh International FLAIRS conference seeks high quality paper
submissions in all areas of AI, including planning, learning, uncertainty
reasoning, computer vision, expert systems, multiagent systems, logic,
knowledge representation, and AI education. All accepted papers will appear
in the conference proceedings, and selected authors will be invited to
submit a full paper to a special issue of the International Journal of
Pattern Recognition and Artificial Intelligence.
SUBMISSIONS
Authors must submit 6 copies of an extended abstract of 1200 to 1600 words.
The extended abstract should not identify the author(s) in any manner.
Please include one separate cover page containing the author name(s),
address, phone number, affiliation, paper title, and topic area. In cases of
multiple authors all correspondence will be sent to the first author unless
otherwise requested.
Abstracts must be received by October 20, 1997.
Abstracts received after this date will not be considered. Notification of
acceptance will be mailed by December 15, 1996. Authors of accepted papers
will be expected to submit the final camera-ready copy of their full papers
by February 23, 1998. Final papers will consist of at most 5 galley pages
(approximately 10 double spaced pages).
For information concerning submissions or to submit an abstract contact:
Diane J. Cook
FLAIRS-98 Program Chair
Box 19015
University of Texas at Arlington
Arlington, TX 76019
Tel: (817) 272-3606
Fax: (817) 272-3784
cook@cse.uta.edu
For general information concerning the conference contact:
Kevin Bowyer or Lawrence O. Hall
FLAIRS-98 General Chairs
Department of Computer Science and Engineering
University of South Florida
4202 E. Fowler Ave
Tampa, FL 33620, USA
Tel: (813) 974-3652
Fax: (813) 974-5456
kwb@csee.usf.edu
hall@csee.usf.edu
THIRD ANNUAL GENETIC PROGRAMMING
CONFERENCE (GP-98)
--------------------------------------------------------------
July 22 - 25 (Wednesday - Saturday), 1998
University of Wisconsin - Madison, Wisconsin
(Held just before AAAI-98 on July 26 - 30, 1998 in Madison)
--------------------------------------------------------------
www.genetic-programming.org
--------------------------------------------------------------
CALL FOR PAPERS AND PARTICIPATION
GENERAL INFORMATION: Genetic programming is an
automatic programming technique for evolving computer
programs that solve (or approximately solve) problems. Over
800 technical papers have been published since 1992 in this
rapidly growing field. The 1997 Genetic Programming
Conference at Stanford University featured 20 tutorials, 3
invited speakers, 69 papers, 15 poster papers in a peer-reviewed
proceedings book published by Morgan Kaufmann Publishers,
as well as vendor presentations and 38 late-breaking papers and
16 PhD student presentations in a separate book. There was a
pre-conference workshop for PhD students. Attendance of GP-
97 was over 350 and exceeded that of the first GP conference in
1996 (288).
TOPICS: Topics include, but are not limited to, applications of
genetic programming, theoretical foundations of genetic
programming, implementation issues, technique extensions, use
of memory and state, cellular encoding (developmental genetic
programming), evolvable hardware, evolvable machine language
programs, automated evolution of program architecture,
evolution and use of mental models, automatic programming of
multi-agent strategies, distributed artificial intelligence,
automated circuit synthesis, automatic programming of cellular
automata, induction, system identification, control, automated
design, compression, image analysis, pattern recognition,
molecular biology applications, grammar induction, and
parallelization.
FOR MORE INFORMATION concerning submitting papers,
hotels, university housing, travel, student travel grants, request
for tutorial proposals, request for workshop proposals, and other
matters, see the GP-98 WWW home page at
For administrative matters, e-mail to
gp@aaai.org
or contact GP-98 Conference, c/o American
Association for Artificial Intelligence, 445 Burgess Drive,
Menlo Park, CA 94025; PHONE: 415-328-3123; FAX: 415-
321-4457. For technical matters, e-mail to John Koza, GP-98
Chair, Computer Science Department, Stanford University at
koza@cs.stanford.edu.