(text)
John Lloyd, CFP: CompulogNet Meeting on Computational Logic
and Machine Learning, June 20th, 1998, Manchester, UK http://www.compulog.org/net-www/MachineLearn.html
--
latest news, publications, tools, meetings, and other relevant items
in the Data Mining and Knowledge Discovery field.
KD Nuggets is currently reaching over 4800 readers in 65+ countries
twice 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 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 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
What I need is a list of specific unknown problems we will encounter.
from a magazine Dilbert quotes contest (thanks to Kathy Wright)
Federal Election Commission Approves White Oak Technologies, Inc. Plan
To Offer Advanced System for Contributor Data Analysis (PR Newswire;
293 words; 04/16/98)
Data Mining System Enhances Campaign and Committee Fundraising While
Targeting Opposition Finance Violations WASHINGTON, April 16
/PRNewswire/ In an Advisory Opinion issued today, the commissioners of
the Federal Election Commission authorized White Oak Technologies,
Inc. (WOTI) to market its package of data mining software and services
to political campaigns and committees. Employing advanced Artificial
Intelligence techniques, WOTI's CampaignMiner(TM) system analyzes
databases to identify hidden patterns of collaboration among people or
organizations
Previous2NextTop
Date: Thu, 02 Apr 1998 09:54:19 +0100
From: Richard Dybowski r.dybowski@umds.ac.uk
Subject: Announcing the ICU-OUTCOME mailing list
*** A New Mailing List: ICU-OUTCOME ***
=======================================
We are announcing the launch of a new mailing list (ICU-OUTCOME) which
focuses on topics related to outcome of patients in intensive care units.
The mailing list is open to doctors, nurses, statisticians, computer
scientists and other allied professionals interested or currently involved
in the study or evaluation of intensive-care related outcome.
Topics for the list include (but are not restricted to) the following:
(a) definitions of outcome (hospital and long-term);
(b) methods for assessing established scoring systems;
(c) design and validation of new prognostic models (statistical and AI-based);
(d) assessment of quality of life;
(e) psychological and social sequelae of critical illness/intensive therapy;
(f) organisational issues around follow-up clinics, funding;
(g) care of physical sequelae, pain management;
(h) rehabilitation, counselling and psychotherapy;
Discussion of related topics is encouraged, including implementation and
description of current practice. We also welcome announcements of
conferences, workshops and study days, as well as postings of current or
future projects and the availability of papers and technical reports.
To subscribe to the mailing list, send an e-mail to majordomo@umds.ac.uk
with the following command in the body of your email message:
subscribe icu-outcome
The mailing list is restricted to those involved with intensive-care
related outcome, whether by research or practise. In order to enable us to
approve your request for membership to the list, it would help if you could
also send a brief e-mail to the list owner (r.dybowski@umds.ac.uk)
describing who you are and what you do. Please include the address of your
Web home page if available.
A Web site for ICU-OUTCOME is under construction.
Dr Alicia Vedio (Intensive Care Research Worker)
Richard Dybowski PhD (Data Analyst and List Owner)
DECISION TECHNOLOGIES FOR FINANCIAL ENGINEERING
Proceedings of the Fourth International Conference on
Neural Networks in the Capital Markets (NNCM'96)
Pasadena, California, USA 20 - 22 November 1996
edited by
Andreas S Weigend (Stern School of Business, New York University),
Yaser S Abu-Mostafa (California Institute of Technology), and
A-Paul N Refenes (London Business School)
(Progress in Neural Processing series, Vol. 7)
This volume selects the best contributions from the Fourth
International Conference on Neural Networks in the Capital
Markets (NNCM). The conference brought together academics
from several disciplines with strategists and decision makers
from the financial industries.
The various chapters present and compare new techniques from
many areas including data mining, information systems,
machine learning, and statistical artificial intelligence.
The volume focuses on evaluating their usefulness for problems
in computational finance and financial engineering.
Applications: Markets:
o risk management o equity
o asset allocation o foreign exchange
o dynamic trading and hedging o bond
o forecasting o commodity
o trading cost control o derivatives
Approaches:
o data mining
o statistical AI
o machine learning
o Monte Carlo simulation
o bootstrapping
o genetic algorithms
o nonparametric methods
o fuzzy logic
The chapters emphasizes in-depth analysis and comparative evaluation
with established approaches.
Readership: Practitioners and academics who are interested in
developments and applications of data mining to finance.
No. of pages: 420pp
Pub. date: Jan 1998
Publisher: World Scientific
ISBN 981-02-3123-7 (hardback) US$96
ISBN 981-02-3124-5 (paperback) US$40
Prices shown are for customers in the US and most of Asia.
Slightly different prices apply in Europe and Japan.
FYI, the NNCM conference has changed its name to 'Computational
Finance', reflecting the expanding set of computational tools
has moved this meeting from its original emphasis on neural
network techniques to a broad selection of different
methodologies. CF99 will be held at NYU/Stern in January 1999,
ARTIFICIAL INTELLIGENCE REVIEW:
ISSUES ON THE APPLICATION OF DATA MINING
<>
Data mining applications vary greatly today and the field can learn
important lessons from this variability. Many important applications
have been developed by using essentially the same data mining
technique. It will be important to understand what type of domain
knowledge or data analysis expertise was used to make such applications
successful. In other successful applications a variety of
complementary techniques had to be used. In such cases it will be
important to understand how the techniques were selected and how the
data was manipulated before it can be mined by each technique, as well
as how the techniques were used cooperatively.
This special issue will highlight some of the current efforts in
applying data mining techniques, with an emphasis on insights that
could help others make the application of those techniques
successful in a real-world situation which is invariably characterized
by large sets of noisy and incomplete data. Of particular interest
would be papers that discuss data mining applications that have been
deployed in production environments or are in the process of being
deployed. Topics could include but are not limited to:
* Issues in data quality, representation, modeling, selection, and
transformation in preparation for mining. Of particular interest is the
relation of these issued to data warehouses and data marts.
* Criteria for selection of a particular data mining technique or sets
of techniques.
* Introduction of additional prior knowledge into the data mining process.
* Integrating a data mining methodology into an existing information
infrastructure.
* Efforts in selecting the most appropriate of the mined knowledge and
in formulating actions based on the mined knowledge.
* Human elements in completing a successful data mining project.
In addition to the call for full-length papers, we request that any
researchers working in this area submit abstracts and/or pointers to
recently published applications for the purpose of compiling a
comprehensive survey of the current state the art.
For Instructions for submitting papers and additional information,
contact the guest editor, or visit Kluwer Academic
Publishers' webpage http://www.wkap.com/.
Papers due: July 1, 1998
Acceptance notification: September 1, 1998
Final manuscript due: Jan 1, 1999
Date of issue: April 1, 1999
Guest Editor:
Julio Ortega
IBM, MS 28-04-4003
1503 LBJ Freeway
Dallas, TX 75234
DBPredictor is a program targeted at on-line classification tasks.
The algorithm uses a lazy model-based approach to focus its effort
on the prediction a single event's class and to return an IF-THEN
based prediction rationale. The specification and source code for
version 2.0 is located at http://www.cs.sfu.ca/~melli/DBPredictor.
You can also directly interact with the program against some
popular datasets. Version 2.0 is the result of my recent Master's
dissertation.
Enhancements from its previous version include:
- dynamic discretization of numeric attributes
- addition of pruning
- ability to tightly-couple with a SQL-based dataset
- support for concept hierarchies
Empirical investigation against 23 datasets suggests that
DBPredictor is:
- generally as accurate as C4.5r8 and IB1(k-NN)
- more accurate than C4.5r8 in the presence of underspecified event
descriptions
- more accurate than IB1 in the presence of irrelevant attributes
- significantly faster than C4.5
PS: I am on the lookout for a new challenge in data mining. Please
contact me if you know of an opportunity that may match my talents,
experience and training. http://www.cs.sfu.ca/~melli/personal/resume.html
Gabor Melli, M.Sc.
School of Computing Science
Simon Fraser University mailto:melli@cs.sfu.ca
Salford Systems presents a seven-city seminar tour, 'Data Mining with
Decision Trees-An Introduction to CART(tm)'. Discover the power of
tree-structured data mining during this one-day course by Dan
Steinberg, a leading expert in CART (classification and regression
tree) technology and real-world applications.
This one-day seminar is geared toward business users and IT audiences
who are interested in understanding CART decision-tree technology and
how to effectively leverage the power of tree-structured data mining
for competitive advantage.
Attendees will learn to:
* Conduct and interpret CART analyses
* Exploit advanced options and controls for more accurate CART models
* Apply CART to make better business decisions
* Improve the predictive accuracy of neural nets and logistic regression
by combining these methods with CART
QUOTES FROM PAST ATTENDEES . . .
'Exceeded my expectations. Superb instructor.'
'Very practical and is easy to understand.'
'Stellar! Has depth and breadth.'
'Excellent overview and explanation.'
SALFORD SYSTEMS' SEVEN-CITY SEMINAR TOUR:
Los Angeles, 4/27/98 * San Francisco, 4/30/98 * Boston, 5/8/98 *
New York, 5/11/98 * Atlanta, 5/13/98 * Dallas, 5/15/98 * Chicago, 5/18/98
Dear knowledge discovery and data mining colleague:
The deadline to submit a proposal for commercial the KDD-98 exhibits is
May 1st, 1998.
I am including the last call for exhibits. This year we are implementing
major changes in the exhibits program. Please take a moment to review
the highlights listed below.
We are looking forward to see you and your products/services in New York
city. For full details see kdd-98-call-for-exhibits.txt
Join Us for KDD-98!
The Fourth International Conference on Knowledge Discovery and Data
Mining (KDD-98) is the premiere event for the data mining community,
bringing together researchers, practitioners and application
developers from such KDD related fields as machine learning,
statistics, databases, data visualization, database marketing and
finance to share ideas/experiences, and to explore new concepts,
applications, tools and techniques. KDD-98 will follow the success of
previous KDD conferences and will feature technical sessions,
tutorials, panels, workshops and exhibits. We invite participants
from universities, industry and government. For more information about
the conference, please visit URL: http://www-aig.jpl.nasa.gov/kdd98
The KDD-98 exhibits program will also follow the success of the
exhibits and demonstration sessions in previous KDD conferences,
featuring demonstrations by the leading vendors of knowledge
discovery and data mining products and services, knowledge discovery
applications and research prototypes. Unlike the previous exhibits,
KDD-98 will:
o Run the exhibits/demos in parallel with the conference so that
schedules do not interfere with each other.
o Make the exhibits/demos separately accessible from the conference
for a nominal fee, less than the conference fee.
o Separate the poster sessions from the exhibits/demos to
differentiate the academic work from the commercial companies.
o Keep the exhibits/demos area open for 1.5 days: 1/2 day on August
28th (Friday) and a full day on August 29th (Saturday).
o Invite the leading consultants providing services in the area of
knowledge discovery and data mining.
o Feature presentations by leading guest speakers in the area of
knowledge discovery and data mining.
o Advertise the exhibits/demos on the internet (in related newsgroups
and electronic journals) and in leading professional journals and
magazines.
All commercial knowledge discovery and data mining tool vendors,
consultants providing related services, academics with research
prototypes, publishers and corporations with significant applications
are welcome to exhibit at KDD-98.
+--------------------------------------------------------------------+
| Important Dates |
+--------------------------------------------------------------------+
The conference will be held in New York City, in midtown Manhattan, at
the New York Marriott Marquis Hotel between August 27-31, 1998. For
more information about the conference, please see www-aig.jpl.nasa.gov/kdd98
The registration deadline for the commercial exhibits is:
May 1, 1998.
If you would like to exhibit at KDD-98, please contact AAAI through
email at:
kdd@aaai.org (also CC: iparsa@epsilon.com)
with the subject line 'KDD-98 Exhibits.' Please send your email before
May 1, 1998 and include the name of your product(s) and/or service(s),
and a 200 word (maximum) description of your product(s) and/or
service(s).
You may also fax or mail this information before May 1, 1998 to:
AAAI
KDD-98 Exhibits
445 Burgess Drive
Menlo Park, CA 94025 USA.
We welcome donations to the conference to specific non-technical
events or to provide give-aways to the attendees. Under some
circumstances, corporate logos can be placed on sponsored items.
General donations are recognized in the following ways:
o Listing of company name
Co-sponsor of KDD-98 for $10,000 or more
'With support from...' for donations under $10,000.
o Signage onsite.
o Brochure and program acknowledgments.
o Listing in the AI Magazine masthead as one AAAI's corporate sponsors.
o One standard exhibit booth space for donations over $2,000.
o One complimentary registration to KDD-98, to be assigned by sponsor.
We also welcome undesignated donations so that they can be used where
the organizers see the most need.
Call for Paper: Workshop: Data Mining in Finance
to be held in conjunction with
The Fourth International Conference on Knowledge Discovery
and Data Mining (KDD 98)
31. August 1998
Marriott Marquis, New York City
Invited speakers: Andreas Weigend, Georg Zimmermann
Submission of papers
The goal of this workshop is to provide an informal forum for researchers
and practitioners to discuss theoretical and applied research issues
of data mining in finance. The topics of interest include, but are not
limited to:
* Are there any special aspects of data mining in finance which are not
typical for data mining in other fields (for examples technical fields,
health, etc.)?
* What are the characteristics of the successful applications of data
mining in finance?
* What are the typical pitfalls?
Application oriented approaches include:
* Volatility models and derivatives pricing
* Risk and liability management
* Portfolio selection and optimization
* Fixed income and term-structure models
* client credit risk and fraud detection
Two panel discussion on efficiency and nolinearity of finance markets
will be organized. Two kind of submissions are solicited: contributed
papers and position statements for panel discussions. All contributed papers
and position statements must be submitted to the following address:
Tae Horn Hann,
Institute for Statistics and Mathematical Economics
University of Karlsruhe
Rechenzentrum, Zirkel 2
76128 Karlsruhe, Germany
e-mail: THH@VWL3SUN1.WIWI.UNI-KARLSRUHE.DE
Phone: 49 721 608 3383
Fax: 49 721 608 3491
Electronic submission (postscript, pdf, or MS Word format) is highly
encouraged. For hard-copy submission, please send three (3) copies of
the full paper to the above address.
TIMETABLE:
Manuscripts due: June 12, 1998
Notification of acceptance/rejection: July 10, 1998
Final version due: July 31, 1998
The First International Conference on Discovery Science
Aqua Plaza Hotel Uminonakamichi, Fukuoka, Japan
December 14-16, 1998
The first international conference on Discovery Science (DS '98) will
be held at Hotel Uminonakamichi, Fukuoka, Japan during December 14 to
16, 1998. The conference will be sponsored by Grant-in-Aid for
Scientific Research on Priority Area ``Discovery Science'' in
cooperation with SIG of Data Mining, Japan Society for Software
Science and Technology.
As we march into the age of digital information, the problem of data
overload looms ominously ahead in almost every field of our society.
Databases of tera byte are now not uncommon. Our ability to analyze
and understand massive datasets lags far behind our ability to gather
and store the data with the ever advancing computer technology. A new
generation of computational techniques and tools is required to
support the extraction and the discovery of useful knowledge from the
rapidly growing volumes of data. Raw data is rarely of direct
benefit. Its true value is reflected by our ability to extract
information useful for decision support or for exploration and
understanding of the phenomena exhibited in the data source.
The ``Discovery Science'' is a three year project from 1998 to 2000
that targets to (1) develop new methods for knowledge discovery, (2)
install network environments for knowledge discovery, and (3)
establish the Discovery Science as a new area of Computer Science. A
systematic research is planned that ranges over philosophy, logic,
reasoning, computational learning and system developments.
The main objective of this conference is to provide an open forum for
intensive discussions and interchange of new information, be it
academic or business, among researchers working in the new area of
Discovery Science.
Topics of interest within the scope of this conference include, but
not limited to, the following areas: Logic for/of knowledge
discovery, knowledge discovery by inferences, knowledge discovery by
learning algorithms, knowledge discovery by heuristic search,
scientific discovery, knowledge discovery in databases, data mining,
knowledge discovery in network environments, inductive logic
programming, abductive reasoning, machine learning, constructive
programming as discovery, intelligent network agents, knowledge
discovery from unstructured and multimedia data, statistical methods
for knowledge discovery, data and knowledge visualization, knowledge
discovery and human interaction, and human factors in knowledge
discovery.
CompulogNet Area Meeting on
Computational Logic and Machine Learning
June 20th, 1998, Manchester, UK
Organiser: John Lloyd, University of Bristol
The next CompulogNet Area Meeting on 'Computational Logic and
Machine Learning' will be held as a Post-Conference Workshop
at JICSLP'98. This meeting is sponsored by the ESPRIT Network of
Excellence in Computational Logic (CompulogNet).
The theme of the meeting will be
'Logic Programming and Machine Learning: A Two-way Connection'.
There is a two-way connection between logic programming and machine
learning. For example, LP has already significantly influenced
(symbolic) ML through the field of inductive logic programming.
There is potential for even greater influence in the near future,
for example, through the application of constraint or higher-order
LP languages, and through the use of abduction. On the other hand,
ML has influenced LP by providing an application area full of
industrially significant problems which can provide a challenge
for the most sophisticated and up-to-date techniques of logic
programming.