*
J. Friedman, Paper On Bias, Variance, 0/1 - Loss, and
the Curse-Of-Dimensionality,
ftp://playfair.stanford.edu/pub/friedman/kdd.ps.Z
*
L. Goldfarb, CFP: Special Issue of Pattern Recognition:
What is inductive learning?
*
X. Wu, CFP: Special Issue of Informatica on Data Mining Metrics
Positions:
*
T. Anand, 2 Positions at Knowledge Discovery group at NCR
Meetings:
*
M. Bramer, Colloquium on Knowledge Discovery, London, October 1996
--
Data Mining and Knowledge Discovery community,
focusing on the latest research and applications.
Contributions are most welcome and should be emailed,
with a DESCRIPTIVE subject line (and a URL, when available) to (kdd@gte.com).
E-mail add/delete requests to (kdd-request@gte.com).
Nuggets frequency is approximately weekly.
Back issues of Nuggets, a catalog of S*i*ftware (data mining tools),
and a wealth of other information on Data Mining and Knowledge Discovery
is available at Knowledge Discovery Mine site, URL http://info.gte.com/~kdd.
-- Gregory Piatetsky-Shapiro (moderator)
********************* Official disclaimer ***********************************
* All opinions expressed herein are those of the writers (or the moderator) *
* and not necessarily of their respective employers (or GTE Laboratories) *
*****************************************************************************
~~~~~~~~~~~~ Quotable Quote ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
THE WALL STREET JOURNAL, May 14, 1996
CALL ITS DUMBSIZING: WHY SOME COMPANIES REGRET COST-CUTTING, p. A1
' ... Despite warnings about downsizing becoming dumbsizing, many companies
continue to make flawed decisions -- hasty, across-the-board cuts -- that come
back to haunt them, on the bottom line, in public relations, in strained
relationships with customers and suppliers, and in demoralized employees.
Sweeping early-retirement and buyout programs sometimes eliminate not only the
deadwood but the talented, many of whom head straight to competitors.'
Previous1NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: Jiawei Han (han@cs.sfu.ca)
Date: Sat, 8 Jun 1996 09:41:21 -0700 (PDT)
Subject: KDD-96 Advance Program
----------------------
KDD'96 Advance Program
----------------------
The Second International Conference on
Knowledge Discovery and Data Mining (KDD-96)
Portland, Oregon, August 2-4 1996.
Sponsored by AAAI and Collocated with AAAI-96 and UAI-96
-----------------------------------------------------------------------
4 invited talks:
o Georges Grinstein, University of Massachusetts at Lowell and MITRE
'Harnessing the Human in Knowledge Discovery,'
o Jeffrey D. Ullman, Stanford University
'Materializing Views in Data Warehouses,'
o Vladimir Vapnik, AT&T Research Laboratories
'Small Sample Size Paradigm in Statistical Inference,'
o Perry K. Youngs, Sara Lee Corporation
'Data Integration and Analysis in a Client Server Environment,'
36 accepted paper presentations.
30 Technology highlight presentations.
3 panel discussions.
2 special sessions and 2 joint KDD and UAI sessions.
+ Poster sessions and data mining system demonstrations.
-----------------------------------------------------------------------
-----------------
Friday August 2
-----------------
8:30-9:45 Plenary session:
Welcome introduction: Evangelos Simoudis (KDD-96 program co-chairman) (15 min).
Invited talk 1: Georges Grinstein (University of Massachusetts at Lowell and
The MITRE Corporation), 'Harnessing the Human in Knowledge Discovery'.
Ronen Feldman (Bar-Ilan Univ., Israel), Haym Hirsh (Rutgers Univ.),
'Mining Associations in Text in the Presence of Background Knowledge'.
Don R. Swanson and Neil R. Smalheiser (Univ. Chicago),
'Undiscovered Public Knowledge: A Ten-Year Update'.
Rakesh Agrawal and Kyuseok Shim (IBM Almaden),
'Developing Tightly-Coupled Data Mining Applications on a Relational
Database System: Methodology and Experience'.
M. Ganesh, Jaideep Srivastava (Univ. Minnesota), and Travis Richardson
(Apertus Technologies, Inc.),
'Mining Entity-Identification Rules for Database Integration'.
George H. John (Stanford Univ.),
'Static versus Dynamic Sampling for Data Mining'.
Session 1 (Plenary session) Scalable Data Mining Systems (40 min: 2 papers)
Gregory Piatetsky-Shapiro (GTE Labs), Ron Brachman (AT&T Research),
Tom Khabaza (ISL, UK), Willi Kloesgen (GMD, Germany), and Evangelos
Simoudis (IBM Almaden), 'An Overview of Issues in Developing Industrial
Data Mining and Knowledge Discovery Applications'.
Paul Stolorz and Christopher Dean (Jet Pro. Lab.), 'Quakefinder:
A Scalable Data Mining System for Detecting Earthquakes from Space'.
Technology spotlight T2 (5 posters, 10 minutes)
Pat Langley (Stanford Univ.),
'Induction of Condensed Determinations'.
Gerald Fahner (UC-Berkeley),
'Data Mining with Sparse and Simplified Interaction Selection'.
Raymond Ng and Edwin Knorr (Univ. British Columbia),
'Extraction of Spatial Proximity Patterns by Concept Generalization'.
Balaji Padmanabhan and Alexander Tuzhilin (New York Univ.),
'Pattern Discovery in Temporal Databases: A Temporal Logic Approach'.
Stephen Mc Kearney (Univ. Bournemouth, UK), Huw Roberts (BT Labs, UK)
'Reverse Engineering Databases for Knowledge Discovery'.
11:00-12:00 Parallel sessions
Session 2A Scalability and Extensibility (60 min: 3 papers)
Stefan Wrobel, Dietrich Wettschereck, Edgar Sommer, and Werner Emde
(GMD, Germany), 'Extensibility in Data Mining Systems'.
Ron Kohavi (Silicon Graphics inc.),
'Scaling Up the Accuracy of Naive-Bayes Classifiers: A Decision-Tree Hybrid'.
Gregory M. Provan (Rockwell) and Moninder Singh (Univ. Pennsylvania),
'Data Mining and Model Simplicity: A Case Study in Diagnosis'.
Session 2B Applications I (60 min: 3 papers)
Jason T. L. Wang (New Jersey Inst. Tech.), Bruce A. Shapiro (US National
Inst. Health), Dennis Shasha (New York Univ.), Kaizhong Zhang (Univ.
Western Ontario, Canada ), Chia-Yo Chang (New Jersey Inst. Tech.),
'Automated Discovery of Active Motifs in Multiple RNA Secondary Structures'
Vic Ciesielski and Greg Palstra (Royal Melbourne Inst. Technology, Australia),
'Using a Hybrid Neural/Expert System for Data Base Mining in Market
Survey Data'.
Shusaku Tsumoto and Hiroshi Tanaka (Tokyo Medical & Dental Univ., Japan),
'Automated Discovery of Medical Expert System Rules from Clinical Databases
based on Rough Sets'.
Invited talk 2: Jeffrey D. Ullman (Stanford University), 'Materializing Views in Data Warehouses'.
2:30-3:30 Plenary sessions
Technology spotlight T3 (5 posters, 10 minutes)
John M. Aronis (Univ. Pittsburgh), Foster J. Provost (NYNEX Science & Tech.),
Bruce G. Buchanan (Univ. Pittsburgh),
'Exploiting Background Knowledge in Automated Discovery'.
David W. Cheung (Univ. Hong Kong), Vincent T. Ng (Hong Kong Polytech. Univ.)
and Benjamin W. Tam (Univ. Hong Kong),
'Maintenance of Discovered Knowledge: A Case in Multi-level Association Rules'.
Arno J. Knobbe and Pieter W. Adriaans (Syllogic, Netherland),
'Analysing Binary Associations'.
Micheline Kamber (Simon Fraser Univ., Canada) and Rajjan Shinghal
(Concordia Univ., Canada), 'Evaluating the Interestingness of
Characteristic Rules'.
Einoshin Suzuki and Masamichi Shimura (Tokyo Inst. Tech., Japan),
'Exceptional Knowledge Discovery in Databases based on Information Theory'.
Session 3 (Plenary session) Spatial and Text Data Mining (40 min: 2 papers)
Martin Ester, Hans-Peter Kriegel, Joerg Sander, and Xiaowei Xu
(Univ. Munich, Germany),
'A Density-Based Algorithm for Discovering Clusters in Large Spatial
Databases with Noise'.
Krista Lagus, Timo Honkela, Samuel Kaski, and Teuvo Kohonen (Helsinki
Univ. Tech., Finland), 'Self-Organizing Maps of Document Collections:
A New Approach to Interactive Exploration'.
Technology spotlight T4 (5 posters, 10 minutes)
David Urpani (Swinburne Univ. Tech., Australia), Xindong Wu (Monash
Univ., Australia), and Jim Sykes (Swinburne Univ. Tech., Australia),
'RITIO - Rule Induction Two In One'.
Kevin J. Cherkauer and Jude W. Shavlik (Univ. Wisconsin-Madison),
'Growing Simpler Decision Trees to Facilitate Knowledge Discovery'.
H.Bodek, R.L.Grossman (Magnify, Inc.), and H.V.Poor (Princeton Univ.),
'Data Mining and Tree-Based Optimization'.
Ron Rymon (Univ. Pittsburgh), 'SE-trees Outperform Decision Trees in
Noisy Domains'.
Pedro Domingos (UC-Irvine), 'Efficient Specific-to-General Rule Induction'.
Session 4A Decision-Tree and Rule Induction (60 min: 3 papers)
Ron Kohavi (Silicon Graphics inc.) and Mehran Sahami (Stanford Univ.),
'Error-Based and Entropy-Based Discretization of Continuous Features'.
Andreas Ittner (Chemnitz Univ. Technology, Germany) and Michael Schlosser
(Fachhochschule Koblenz, Germany), 'Discovery of Relevant New Features
By Generating Non-Linear Decision Trees'.
Pedro Domingos (UC-Irvine), 'Linear-Time Rule Induction'.
Session 4B (Special session) Systems for Mining Large Databases (60 min: 3 papers)
Rakesh Agrawal, Andreas Arning, Toni Bollinger, Manish Mehta, John Shafer,
R. Srikant (IBM Almaden Research Center), 'The Quest Data Mining System'.
Tomasz Imielinski and Aashu Virmani (Rutgers Univ.),
'DataMine - Application Programming Interface and Query Language for
KDD applications'.
Jiawei Han, Yongjian Fu, Wei Wang, Jenny Chiang, Wan Gong, Krzysztof
Koperski, Deyi Li, Yijun Lu, Amynmohamed Rajan, Nebojsa Stefanovic,
Betty Xia, and Osmar R. Zaiane (Simon Fraser Univ.),
'DBMiner: A System for Mining Knowledge in Large Relational Databases'
4:50-5:30 Parallel sessions
Session 5A Mining with Noise and Missing data (40 min: 2 papers)
Kamakshi Lakshminarayan, Steve Harp, Robert Goldman, and Tariq Samad
(Honeywell), 'Imputation of Missing Data Using Machine Learning Techniques'.
Heikki Mannila and Hannu Toivonen (Univ. Helsinki, Finland),
'Discovering generalized episodes using minimal occurrences'
Session 5B Panel Discussion: Systems for Mining Large Databases (40 min)
----------------------------------------
6:00-8:00 Opening reception
----------------------------------------
(in parallel with: The Poster and Demonstration Session).
-----------------
Saturday August 3
-----------------
8:30-9:30 Plenary session:
Invited talk 3: Vladimir Vapnik (AT&T Research Laboratories),
'Small Sample Size Paradigm in Statistical Inference'.
Ian W. Flockhart (Quadstone Ltd., UK), and Nicholas J. Radcliffe
(Univ. Edinburgh & Quadstone Ltd, UK), 'A Genetic Algorithm-Based
Approach to Data Mining'.
Tae-Wan Ryu and Christoph F. Eick (Univ. Houston), 'Deriving Queries
>from Results using Genetic Programming'.
Ning Shan, Wojciech Ziarko, Howard J. Hamilton, and Nick Cercone
(Univ. Regina, Canada), 'Searching Classification Knowledge in Databases
Based on Rough Sets'.
Yang Wang and Andrew K.C.Wong (Univ. Waterloo, Canada),
'Representing Discovered Patterns Using Attributed Hypergraph'.
Takao TERANO and Yoko ISHINO (Univ. Tsukuba, Japan),
'Interactive Knowledge Discovery from Marketing Questionnaire Using
Simulated Breeding and Inductive Learning Methods'.
Alvaro Monge and Charles Elkan (UC-San Diego),
'The field matching problem: Algorithms and applications'.
M. Richeldi (CSELT S.p.A., Italy) and P.L. Lanzi (Politecnico di
Milano, Italy), 'Performing Effective Feature Selection by
Investigating the Deep Structure of the Data'.
Thomas Hofmann and Joachim Buhmann (Univ. Bonn, Germany), 'Inferring
Hierarchical Clustering Structures by Deterministic Annealing'.
Stefan Kramer and Bernhard Pfahringer (Austrian Research Inst. for
Artificial Intelligence, Austria),
'Efficient Search for Strong Partial Determinations'.
11:05-12:05 Parallel sessions
Session 7A Prediction (60 min: 3 papers)
Petri Kontkanen, Petri Myllymaki, and Henry Tirri (Univ. Helsinki, Finland),
'Predictive Data Mining with Finite Mixtures'.
Rense Lange (Univ. Illinois-Springfield),
'An Empirical Test of the Weighted Effect Approach to Generalized Prediction
Using Recursive Neural Nets'
Robert Engels (Univ. Karlsruhe, Germany),
'Planning Tasks for Knowledge Discovery in Databases; Performing
Task-Oriented User-Guidance'.
Session 7B Applications II (60 min: 3 papers)
Usama Fayyad (Microsoft Research), David Haussler (UC-Santa Cruz), and
Paul Stolorz (Jet Prop. Lab.), 'KDD for Science Data Analysis: Issues
and Examples'.
Ruediger Wirth and Thomas P. Reinartz (Daimler-Benz Res. & Tech.,
Germany),
'Detecting Early Indicator Cars in an Automotive Database: A
Multi-Strategy
Approach'.
B. de la Iglesia, J.C.W. Debuse, and Rayward-Smith V.J. (Univ. East
Anglia),
'Discovering Knowledge in Commercial Databases Using Modern Heuristic
Techniques'.
Invited talk 4: Perry K. Youngs (Sara Lee Corporation),
'Data Integration and Analysis in a Client Server Environment: The Sara Lee
Meat Experience'.
2:30-3:30 Plenary session
Session 8 (Plenary session) Combining Data Mining and Machine Learning (60 min: 3 papers)
Tom Fawcett and Foster Provost (NYNEX Science and Technology),
'Combining Data Mining and Machine Learning for Effective User Profiling'.
Philip Chan (Florida Inst. Tech.), and Salvatore Stolfo (Columbia Univ.),
'Sharing Learned Models among Remote Database Partitions by Local
Meta-Learning'.
Truxton Fulton, Simon Kasif, and Steven Salzberg (Johns Hopkins Univ.),
'Local Induction of Decision Trees: Towards Interactive Data Mining'.
Session 9A Approaches to Numeric Data (60 min: 3 papers)
Andrzej Czyzewski (Tech. Univ. Gdansk, Poland), 'Mining Knowledge in
Noisy Audio Data'.
Kenneth A. Kaufman and Ryszard S. Michalski (George Mason Univ.), 'A
Method for Reasoning with Structured and Continuous Attributes in the
INLEN-2 Knowledge Discovery System'.
A.J. Feelders (Data Distilleries),
'Learning from Biased Data Using Mixture Models'.
Session 9B (Special Session) Scalable and distributed applications of
KDD (60 min: 3 papers)
D. Pfitzner and J. Salmon (Mount Stromlo Observatory and Caltech.),
'Halo Finding in N-body Cosmology Simulations on Massively Parallel
Computers'.
E. Mesrobian, R. Muntz, E. Shek and K. Ng (UCLA),
'Scalable Exploratory Data Mining on Distributed Geoscientific Data'.
I. Hovacker, M. Huynen, P. Stadler and P. Stolorz, (U. of Illinois,
LANL, U. Vienna and JPL), 'Knowledge Discovery in RNA Sequence Families
of HIV Using Scalable Computers'.
4:50-5:30 Parallel sessions
Session 10A Pattern-Oriented Data Mining (40 min: 2 papers)
Wei-Min Shen and Bing Leng (USC),
'Metapattern Generation for Integrated Data Mining'.
Jan Zytkow and Robert Zembowicz (Wichita State Univ.),
'Automated Pattern Mining with a Scale Dimension'.
Session 10B Panel discussion: Scalable and Distributed Applications of KDD -
The Promise and Challenge of Data Mining with High Performance Computers
(40 min: 2 papers)
8:30-11:35 Sessions 11-12 Joint UAI-KDD plenary sessions
Selected talks on learning graphical models from the UAI and
KDD proceedings. UAI badges will be honored at the Portland
Convention Center for the joint session.
8:30-8:40 Introductory remarks for joint session: UAI meets KDD
Usama Fayyad and Eric Horvitz (Microsoft Research)
[KDD]: Usama Fayyad (Microsoft Research), Gregory Piatetsky-Shapiro,
(GTE Labs), and Padhraic Smyth (Jet Prop. Lab.),
'Knowledge Discovery and Data Mining: Toward a Unifying Framework'.
[UAI]: D. Chickering (UCLA) and D. Heckerman (Microsoft Research),
'Efficient Approximations for the Marginal Likelihood of Incomplete
Data Given a Bayesian Network'.
[KDD]: Padhraic Smyth (Jet Propulsion Lab.), 'Clustering using Monte
Carlo Cross-Validation'
[UAI]: D. Chickering (UCLA), 'Learning Equivalence Classes of Bayesian
Network Structures'.
----------------------------------------
10:00-10:15 AM Coffee Break
----------------------------------------
[UAI]: N. Friedman (Stanford Univ.) M. Goldszmidt (SRI), Learning
Bayesian Networks with Local Structure.
[KDD]: Ron Musick (Lawrence Livermore Nat. Lab.), 'Rethinking the
Learning of Belief Network Probabilities'.
[UAI]: K. Laskey and L. Martignon , 'Bayesian Learning of Loglinear
Models for Neural Connectivity'.
[KDD]: Paul Stolorz ((Jet Prop. Lab.) and Philip Chew (U. Penn),
'Harnessing Graphical Structure in Markov Chain Monte Carlo Learning'.
----------------------------------------
11:35-12:30 Lunch will be served.
(which may partially overlap with the Summary Panel Session).
----------------------------------------
12:30-2:00 PM Plenary Session and Final Wrap-up Meeting
12:30 - 1:20 PM Summary panel and closing remarks: 'What Have We Discovered?'
1:20 - 2PM KDD Wrap-up Business Meeting:
Plans for the future, feedback from attendees, issues of general
interest to the KDD Community as a whole.
2PM - Conference Adjourns -- Start of AAAI Workshops and Tutorials
Previous2NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Wed, 5 Jun 1996 16:36:51 -0400 (EDT)
From: Patrick Hoffman (phoffman@cs.uml.edu)
Subject: Information Exploration Shootout
Previous3NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: tanand@winhitc.Atlantaga.ncr.com
(Tej Anand)
Subject: List of KDD-96 Demos
The following is a list of demos for KDD-96. All demo proposals
that I received by the June 3rd deadline were accepted.
-Tej.
------------------------------------------------------------------
=
Kepler: Extensibility in Data Mining Systems
Stefan Wrobel, Dietrich Wettschereck, Edgar Sommer, Werner Emde
GMD, FIT.KI, Schloss Birlinghoven, 53754 Sankt Augustin, Germany
(Stefan.Wrobel@gmd.de)
=
DBMiner: A System for Mining Knowledge in Large Relational Databases=
Jiawei Han, Yongjian Fu, Wei Wang, Jenny Chiang, Wan Gong, Krzysztof=
Koperski, Deyi Li, Yijun Lu, Amynmohamed Rajan, Nebojsa Stefanovic, =
Betty Xia, Osmar R. Zaiane.
School of Computing Science, Simon Fraser University, British =
Columbia, Canada V5A 1S6
(han@cs.sfu.ca)
=
WebFind: Mining external sources to guide WWW discovery.
Alvaro E. Monge and Charles P. Elkan
Department of Computer Science and Engineering University of =
California, San Diego, La Jolla, CA 92093-0114
{amonge,elkan}@cs.ucsd.edu
=
MM -- Mining with Maps
Raymond Ng
Department of Computer Science, University of British Columbia
(rng@cs.ubc.ca)
=
Decisionhouse
Nick Radcliffe
Quadstone Ltd., UK
(njr@quadstone.co.uk)
=
STARC - A New Data Mining Tool
Damir Gainanow, Andre Matweew, Michael Thess
DATA-Center Ltd., Ekaterinburg, RUSSIA
Scholz & Thess Software GbR, Chemnitz, GERMANY
(michael.thess@Mathematik.TU-Chemnitz.DE)
=
D-SIDE: A probabilistic DeciSIon enDorsement Environment
P.Kontkanen, P.Myllym=E4ki and H.Tirri
Complex Systems Computation Group, Department of Computer Science
University of Helsinki, Finland
(myllymak@cs.Helsinki.FI)
=
MineSet
Steven Reiss, Mario Schkolnick =
Data Mining and Visualization Group, Silicon Graphics Computer Syste=
ms
(sreiss@powerplay.engr.sgi.com)
=
Optimization Related Data Mining using the PATTERN System
H. Bodek, R. L. Grossman, D. Northcutt, H. V. Poor
Magnify, Inc.
Princeton University
(rlg@opr.com)
=
Management Discovery Tool
Ken O'Flaherty
NCR Corporation
(ken.oflaherty@sandiegoca.ncr.com)
=
Clementine Data Mining System
Colin Shearer
Data Mining Division, Integral Solutrions Ltd
colin@isl.co.uk
=
Mining Associations in Text in the Presence of Background Knowledge
Ronen Feldman, Haym Hirsh
Bar-Ilan Univesity
Rutgers University.
(feldman@cs.biu.ac.il)
=
=
DataMine: Ad Hoc KDD querying
Tomasz Imielinski and Aashu Virmani
Rutgers University
(avirmani@paul.rutgers.edu)
= Previous4NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: Usama Fayyad (fayyad@MICROSOFT.com)
Subject: FW: CDT on Internet privacy policy
Date: Fri, 31 May 1996 09:11:38 -0700
[CDT's http://www.cdt.org/privacy/
page includes information on
tracking Web use (e.g. cookies, server logs, etc.), the privacy
policies of AOL, MSN, CompuServe, and Prodigy, and anonymous Web
browsing.
=================
From editor@cdt.org
Thu May 30 21:46:26 1996
Subject: Policy Post 2.21 - Your Privacy Online: CDT Unveils Demo &
Clearinghouse
CONTENTS: (1) Your Privacy Online - CDT Unveils Demonstration and
Clearinghouse
(2) Join Rep. White Wed 6/5 At HotWired to Discuss the
Internet
Caucus, the CDA, and other Internet Policy Issues
(3) Subscription Information
(4) About CDT, contacting us
** This document may be redistributed freely with this banner in tact
**
Excerpts may be re-posted with permission of (editor@cdt.org)
------------------------------------------------------------------------
-----
(1) Your Privacy Online - CDT Unveils Demonstration & Clearinghouse
Many people surf the World Wide Web with an illusion of anonymity,
believing that their activities are unobserved and that they can
explore
the Internet without leaving a trail. In reality, this is not the
case.
During the normal course of using the Internet, a great deal of
personally
revealing information is routinely generated, collected, and stored.
Most
of this information is collected for purposes of system maintenance,
billing, or other necessary functions. But a sophisticated marketer,
determined hacker, or law enforcement official can put together a
detailed
profile of your online activities, personal tastes, interests, habits
and
vices with relative ease.
Today, the Center for Democracy and Technology unveiled an interactive
privacy demonstration and privacy policy clearinghouse on our World
Wide
Web site. The demonstration is located at
1. To educate the public about the extent to which personal information
is automatically revealed online, and
2. To begin to make available examples of privacy and information usage
policies that give people greater knowledge of and control over the
personal information revealed online. The current focus of this
'Privacy Clearinghouse' is centered on commercial online service
providers (ie, America Online, CompuServe, Prodigy, Microsoft).
Future updates of the clearinghouse will include information on
Internet
Service Providers, content providers, and web browser software. Future
updates will also explore the extent that users can employ various
technological solutions to control the collection and disclosure of
personal
information.
FEDERAL TRADE COMMISSION TO HOLD HEARINGS ON ONLINE PRIVACY
On June 4 and 5, the Federal Trade Commission (FTC) will hold hearings
to
explore online privacy issues. The FTC is particularly interested in
exploring privacy protecting technologies which empower users to
exercise
more control over the collection and use of personally identifiable
information online.
CDT has been invited to present testimony at the hearings. Testimony
and
other background information on the FTC hearings will be available at
CDT's
web page at the end of next week. Details on the hearings are available
at
http://www.ftc.gov/
WHY SHOULD NETIZENS CARE ABOUT THIS ISSUE?
Although it may not seem like it, someone is following you through
cyberspace. Every time you retrieve a file, view an image, send an
email
message or jump to a new web site, a record is created somewhere on the
Net.
While much of this information may never be used, it can be, and you
have
little control over it. In the hands of a marketer with a powerful
computer, or the government, it is possible to build a detailed profile
of
your tastes and preferences by monitoring your online activities.
The information can be used to send you unsolicited email or snail
mail, to
call you, or to even put you on a list of people likely to support a
particular political candidate. A single piece of information about
you
can support a tremendous range of activities. For example, if your
repeated visits to web sites containing information on cigarettes
results
in free samples, coupons, or even email to you about a new tobacco
product,
you may not be concerned. However, if your visits to these web sites
result in escalating insurance premiums due to categorization as a
smoker -
now you're beginning to get concerned.
HOW PERSONAL INFORMATION IS COLLECTED ONLINE
Web sites and Web browsers
--------------------------
Your personal information (including your hobbies, political and
product
interests and ways to contact you, such as your email address) can be
collected by web sites in two ways: directly or indirectly.
* PASSIVE RECORDING OF TRANSACTIONAL INFORMATION: The transactional
information revealed in the normal course of surfing the net reveals a
great deal of information about your online activities. When you
visit
a particular web site, for example, the webmaster can determine what
files, pictures, or other information you are most interested in (and
what you ignored), how long you examined a particular page, image or
file, where you came from, where you went to.
Web servers collect transactional information in order to allow the
system operator to perform necessary system maintenance, auditing,
and
other essential system functions. However, when correlated with
other
sources of personal information, including marketing databases, phone
books, voter registration lists, etc, a detailed profile of your
online activities can be created without your knowledge or consent.
* COOKIES: Additionally, many web browsers contain a feature called
'cookies,' or client-side persistent information. Cookies allows any
web site to store information about your visit to that site on your
hard drive. Every time you return to that site, 'cookies' will read
your hard drive to find out if you've been there before. (The Privacy
Demonstration has a link to a site that utilizes cookies.)
* DIRECT DISCLOSURE OF PERSONAL INFORMATION: A growing number of web
sites offer users the ability to register with the site. In many
cases, registration brings real, important benefits, such as access
to
special areas, timely information, discounts, etc. While
registration
or other mechanisms by which users divulge personal information to a
web site provide some obvious benefits to a users, it also provides
the site's operator with a detailed picture of how you use the site.
Regardless of how the information is obtained, a great deal of
personally
identifiable information is revealed in the normal course of surfing
the
web.
Commercial Online Service Providers
-----------------------------------
Commercial online service providers are configured in a variety of
ways,
but generally, little personally identifiable information is revealed
to
Internet sites visited directly from an online service.
If you subscribe to a commercial online service, your service provider
has
access to lots of information about your online activities. These
records
are generated in the normal course of using the service, and are
important
for billing and maintenance purposes. However, not all services treat
the
use and disclosure of this information the same way.
Please visit The Center for Democracy and Technology's Clearinghouse on
Privacy Policies http://www.cdt.org/privacy/
for a detailed
description of
the information practices of the major commercial online services.
Future updates of the clearinghouse will focus on other Internet
entities,
such as browsers, content providers, and Internet service providers.
(2) JOIN CONGRESSMAN RICK WHITE (R-WA) LIVE ONLINE TO TALK ABOUT THE
INTERNET CAUCUS, THE CDA, AND TAKE YOUR QUESTIONS
Congressman Rick White (R-WA) will be live online at HotWired on
Wednesday
June 5 at 9:00 pm ET to discuss his efforts to encourage better
communication between members of Congress and the Internet community,
his
plans for the Congressional Internet Caucus, and other topics.
Representative White will also answer questions from Netizens.
DETAILS ON THE EVENT
* Wednesday June 5, 9 - 10 pm ET (6 pm Pacific) on HotWired
To participate, you must be a registered HotWired member (there
is no charge for registration). You must also have RealAudio(tm) and
a telnet application properly configured to work with your browser.
Wednesday's forum is another in a series of planned events, and is part
of a broader project coordinated by CDT and the Voters
Telecommunications
Watch (VTW) designed to bring the Internet Community into the debate
and
encourage members of Congress to work with the Net.community on vital
Internet policy issues.
Transcripts from last week's discussion with Senator Leahy are
available at http://www.cdt.org/crypto/.
Events with other members of Congress
working
on Internet Policy Issues are currently being planned. Please check http://www.cdt.org/
for announcements of future events
Be sure you are up to date on the latest public policy issues affecting
civil liberties online and how they will affect you! Subscribe to the
CDT
Policy Post news distribution list. CDT Policy Posts, the regular news
publication of the Center For Democracy and Technology, are received by
more than 9,000 Internet users, industry leaders, policy makers and
activists, and have become the leading source for information about
critical free speech and privacy issues affecting the Internet and
other
interactive communications media.
To subscribe to CDT's Policy Post list, send mail to
policy-posts-request@cdt.org
with a subject:
subscribe policy-posts
If you ever wish to remove yourself from the list, send mail to the
above address with a subject of:
unsubscribe policy-posts
-----------------------------------------------------------------------
(4) ABOUT THE CENTER FOR DEMOCRACY AND TECHNOLOGY/CONTACTING US
The Center for Democracy and Technology is a non-profit public interest
organization based in Washington, DC. The Center's mission is to
develop
and advocate public policies that advance democratic values and
constitutional civil liberties in new computer and communications
technologies.
Contacting us:
General information: info@cdt.org
World Wide Web: http://www.cdt.org/
FTP ftp://ftp.cdt.org/pub/cdt/
Snail Mail: The Center for Democracy and Technology
1634 Eye Street NW * Suite 1100 * Washington, DC 20006
(v) +1.202.637.9800 * (f) +1.202.637.0968
-----------------------------------------------------------------------
End Policy Post 2.21 5/30/96
----------------------------------------------------------------------- Previous5NextTop
>~~~Publications:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: 'Jerome H. Friedman' (jhf@playfair.Stanford.EDU)
To: kdd@eureka
Subject: Paper Announcement.
ON BIAS, VARIANCE, 0/1 - LOSS, AND
THE CURSE-OF-DIMENSIONALITY
Jerome H. Friedman
Stanford University
(jhf@playfair.stanford.edu)
ABSTRACT
The classification problem is considered in which an output variable
assumes discrete values with respective probabilities that depend upon the
simultaneous values of a set of input variables. At issue is how error in
the estimates of these probabilities affects classification error when the
estimates are used in a classification rule. These effects are seen to be
somewhat counter intuitive in both their strength and nature. Rather than
being additive, the bias and variance components of the estimation error
are shown to exhibit a strong multiplicative interaction effect on
classification error. Certain types of (very high) bias can be canceled by
low variance to produce accurate classification. This causes much of the
intuition derived from estimation theory to be misleading when applied to the
classification problem. It also explains many of the counter intuitive aspects
of classification that have been observed empirically. For example, it helps
explain why simple methods like 'naive' Bayes and nearest neighbors are
often competitive with and sometimes superior to more sophisticated ones
for classification, and why 'bagging/aggregating' classifiers can often
improve accuracy. These results also suggest simple modifications to these
procedures that can (sometimes dramatically) further improve their
classification performance. It is also shown that the effect of the curse-
of-dimensionality can be much less severe for classification than for function
estimation.
Available by ftp from:
ftp://playfair.stanford.edu/pub/friedman/kdd.ps.Z
Previous6NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Sun, 9 Jun 1996 23:21:00 -0300 (ADT)
From: Lev Goldfarb (goldfarb@unb.ca)
Reply-To: Lev Goldfarb (goldfarb@unb.ca)
Subject: Call for papers for a Special Issue of Pattern Recognition: What is inductive learning?
Call for papers
-----------------
Special Issue of Pattern Recognition
(The Journal of the Pattern Recognition Society)
WHAT IS INDUCTIVE LEARNING:
ON THE FOUNDATIONS OF PATTERN RECOGNITION, AI, AND COGNITIVE SCIENCE
Guest editor: Lev Goldfarb
Faculty of Computer Science
University of New Brunswick
Fredericton, N.B., Canada
The 'shape' of AI (and, partly, of cognitive science), as it stands now,
has been molded largely by the three founding schools (at Massachusetts
institute of Technology, Carnegie-Mellon and Stanford Universities). This
'shape' stands now fragmented into several ill-defined research agendas
with no clear basic SCIENTIFIC problems (in the classical understanding of
the term) as their focus. It appears that four factors have contributed to
this situation: inability to focus on the central cognitive process(es),
lack of understanding of the structure of advanced scientific models,
failure to see the distinction between the computational/logical and
mathematical models, and the relative abundance of research funds for AI
during the last 35 years.
The resulting research agendas have prevented AI from cooperatively
evolving into a scientific discipline with some central 'real' problems
that are inspired by the basic cognitive/biological processes. The
candidates for such basic processes could come only from the
central/common perceptual processes and only much later employed by the
'higher', e.g. language, processes: the period during which the 'higher'
level processes have evolved is insignificant compared to that in which
the development of the perceptual processes took place (compare also the
anatomical development of the brain which does not show any basic changes
with the development of the 'higher' processes).
Moreover, the partisan tradition in the development of AI may have also
inspired the recent 'connectionist revolution' as well as other smaller
'revolutions', e.g., that related to the 'genetic' learning. As a result,
in particular, even the most 'reputable' connectionist histories of the
field of pattern recognition, which was formed more than three decades ago
and to which the connectionism properly belongs, show amazing ignorance of
the major developments in the parent (pattern recognition) field: the
emergence of two important and formally quite irreconcilable recognition
paradigms--vector space and syntactic. The latter ignorance is even more
instructive in view of the fact that many engineers who got involved with
the field of pattern recognition through the connectionist 'movement' are
also ignorant of the above two paradigms that were discovered and
developed within largely applied/engineering parent field of pattern
recognition.
As far as the inception of a scientific field is concerned, it should be
quite clear that the initial choice of the basic scientific problem(s) is
of decisive importance. This is particularly true for cognitive modeling
where the path from the model to the experiment and the reverse path are
much more complex than was the case, for example, at the inception of
physics. In this connection, a very important question arises, which will
be addressed in the special issue: What form will the future/adequate
cognitive models take?
Furthermore, may be, as many cognitive scientists argue, since we are in a
prescientific stage, we should simply continue to collect more and more
data and not worry about the future models. The answer to the last
argument is quite clear to me: look very carefully at the 'data' and the
corresponding experiments and you will note that no data can be even
collected without an underlying model, which always includes both formal
and informal components. In other words, we cannot avoid models
(especially in cognitive science, where the path from the model to the
experiment will be much longer and more complex than is the case in all
other sciences). Therefore, paraphrasing Friedrich Engels's thought on the
role of philosophy in science, one can say that there is absolutely no way
to do a scientific experiment without the underlying model and the
difference between a good scientist and a bad one has to do with the
degree to which each realizes this dependence and actively participates in
the selection of the corresponding model. It goes without saying that, at
the inception of the science, the decision on which cognitive process one
must focus initially should precede the selection of the model for the
process.
As to the choice of the basic scientific problem, or basic cognitive
process, it appears that the really central cognitive process is that of
inductive learning, which might have been marked so by many great
philosophers of the past four centuries (e.g., Bacon, Descartes, Pascal,
Locke, Hume, Kant, Mill, Russell, Quine) and even earlier (e.g.,
Aristotle). The insistence of such outstanding physiologists and
neurophysiologists as Helmholtz and Barlow on the central role of
inductive learning processes is also well known. However, in view of the
difficulties associated with developing an adequate inductive learning
model, researchers in AI and to a somewhat lesser extent in cognitive
science have decided to view inductive learning not as a central process
at all, i.e., they decided to 'dissolve' the problem.
It became clear to me that the above difficulties are related to the
development of a genuinely new (symbolic) mathematical framework that can
SATISFACTORILY define the concept of INDUCTIVE CLASS REPRESENTATION (ICR),
i.e., the nature of encoding essentially infinite data set on the basis of
a small finite set. (The most known as well as critical to the development
of mathematics example of ICR is that of the classical Peano
representation of the set of natural numbers--one element plus one
operation--used in mathematical induction.) Thus, the main differences
between inductive learning models should be viewed in light of the
differences between the formal means, i.e. mathematical structures,
offered by various models for representing the class inductively. I will
also argue (in one of the papers) that the classical mathematical
(numeric) models, including the vector space and probabilistic models,
offer inadequate axiomatic frameworks for capturing the concept of ICR.
As Peter Gardenfors aptly remarked in his 1990 paper, 'induction has been
called 'the scandal of philosophy' [and] unless more consideration is
given to the question of which form of knowledge representation is
appropriate for mechanized inductive inferences, I'm afraid that induction
may become a scandal of AI as well.' I strongly believe that all attempts
to 'dissolve' the inductive learning processes are futile and, moreover,
that these processes are central cognitive processes for all levels of
processing, hence the earlier workshop in Toronto (May 20-21) under the
same title and the present Special Issue.
I invite all researchers seriously interested in the scientific
foundations of cognitive science, AI, or pattern recognition to submit
the papers addressing, in addition to other relevant issues, the following
questions:
* What is the role of mathematics in cognitive science, AI,
and pattern recognition?
* Are there any central cognitive processes?
* What is inductive learning?
* What is inductive class representation (ICR)?
* Are there several basic inductive learning processes?
* Are the inductive learning processes central?
* What are the relations between inductive learning
processes and the known physical processes?
* What is the relationship between the measurement
processes and inductive learning processes (e.g., retina
as a structured measurement device)?
* What is the role of inductive learning in sensation
and perception (vision, hearing, etc.)?
* What is the relation between the inductive learning,
categorization, and pattern recognition?
* What is the relation between the supervised/inductive
learning and the unsupervised learning?
* What is the role of inductive learning processes in
language acquisition?
* What are the relationships, if any, between the inductive
class representation (ICR) and the basic object
representation (from the class)?
* What are the differences between the mathematical
structures employed by the known inductive learning
models for capturing the corresponding ICRs?
* What is the role of inductive learning in memory and
knowledge representation?
* What are the relations, if any, between the ICR and
mental models and frames?
When preparing the manuscript, please conform to the standard submission
requirements given in journal Pattern Recognition, which could be faxed or
mailed if necessary. Hardcopies (4) of each submission should be mailed
to
Lev Goldfarb
Faculty of Computer Science
University of New Brunswick
P.O. Box 4400 E-mail: goldfarb@unb.ca
Fredericton, N.B. E3B 5A3 Tel: 506-453-4566
Canada Fax: 506-453-3566
by the SUBMISSION DEADLINE, August 20, 1996. The review process should
take about 4-5 weeks and will take into account the relevance, quality,
and originality of the contribution. Potential contributors are encouraged
to contact me with any questions they might have.
Previous7NextTop
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: xindong@insect.sd.monash.edu.au
Date: Wed, 12 Jun 1996 13:42:57 +1000 (EST)
Subject: CFP: Special Issue of Informatica on Data Mining Metrics
CFP: A SPECIAL ISSUE OF INFORMATICA ON DATA MINING METRICS
The development of a large number of rule induction and decision tree
construction algorithms for data mining by researchers in machine
learning and statistics, has seen empirical evaluation and
justification become an important aspect for acceptance of newly
developed algorithms by researchers in the field. To provide a
comprehensive evaluation, we need a set of standard criteria such as:
induction time, size of induction results, time to execute the
induction results, and predicative accuracy. One algorithm may be
able to perform better than others with one criterion, but may perform
poorly with other criteria. With the same set of algorithms, we can
also get different evaluation results from different sets of
databases. The question of why, and under which circumstances one
algorithm (whether it is newly designed or an existing one)
outperforms others becomes more important than simply presenting
empirical results from an arbitrarily selected set of databases.
Research on data mining metrics is based on the above mentioned,
widely adopted criteria. These metrics also look into the
characteristics of the data sets for experiments such as: the numbers
of classes, attributes and examples, the distribution of training
examples in the example space, the level of noise and the mixture of
continuous and nominal values. The aim is to develop a meaningful set
of metrics with well documented experiment results for different
algorithms. These metrics can be used as a testbed for newly developed
algorithms against existing ones.
Original papers are solicited that describe research in data mining
metrics for a special issue in Informatica: An International Journal
of Computing and Informatics. Topics include, but are not limited to,
the following:
- Development of data mining metrics
- Different knowledge representations and their transformations in
data mining
- Generality and complexity of induction results
- Pre-pruning and post-pruning
- Deduction of inexact induction results
- Artificial databases and real world databases: Which are more
suitable for experimentation?
- Top-down vs. bottom-up induction algorithms
- Discretisation of real-valued attributes and fuzzification of
symbolic values for data mining
TIMETABLE
Papers in 5 hard copies due: 12 August 1996
Acceptance Notification: 15 October 1996
Final papers in LaTeX: 15 November 1996
Publication: December 1996
GUEST EDITORS
Dr Xindong Wu (xindong@insect.sd.monash.edu.au)
Department of Software Development, Monash University
900 Dandenong Road, Caulfield East, Melbourne 3145, Australia
URL: http://www.sd.monash.edu.au/~xindong/
Phone: +61 3 9903 1025 Fax: +61 3 9903 1077
Dr Matjaz Gams (matjaz.gams@ijs.si)
Jozef Stefan Institute, Intelligent Systems Department
Jamova 39, 61000 Ljubljana, Slovenia
URL: http://www2.ijs.si/~mezi/matjaz.html
Phone: +386 61 1773 900 Fax: +386 61 1258 058
Previous8NextTop
>~~~Positions:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Date: Tue, 11 Jun 1996 10:22:35 -0400
From: tanand@winhitc.Atlantaga.ncr.com
(Tej Anand)
Subject: Job posting for KDD Nuggets
----------------------------------------------------------------------
We have 2 immediate openings in the Knowledge Discovery group at NCR's
Human Interface Technology Center (HITC).
Both these positions require at least a graduate degree in Computer
Science with specialization in machine learning, statistics or very
large databases. Work within a specialization should demonstrate
experience in dealing with issues pertaining to knowledge discovery.
NCR is the world leader for commercial data warehouse solutions.
NCR's customers include some of the world's largest retail, consumer
packaged goods, transportation (cargo & passenger),
telecommunications, insurance and banking enterprises.
The HITC, located in midtown Atlanta, is an R&D group of approximately
100 individuals. HITC's vision is to revolutionize the way computer
based products relate to people. Researchers at HITC have
unprecedented access to NCR customers. Customer partnerships give us
a context and focus to address challenging problems in the area of
human-computer interaction. Our core competencies include user
centered design, multi-modal user interfaces, mobile computing,
audio/video processing, synthetic environments, image understanding,
case-based reasoning and knowledge discovery.
The Knowledge Discovery team at the HITC was established in 1993 to
enable commercial enterprises realize business insights hidden in
their operational data. This team conducts R&D to develop innovative
knowledge discovery processes and data mining tools. All R&D is
conducted in partnership with NCR customers.
To apply for the openings send your resume via e-mail to
tej.anand@atlantaga.ncr.com
or by regular mail to
Tej Anand
NCR Corporation
Prom I, Room 8147
1200 Peachtree St.
Atlanta, GA 30309.
Previous9NextTop
>~~~Meetings:~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
From: bramerma@cv.port.ac.uk
Date: Sat, 15 Jun 1996 18:57:53 EDT
Subject: Colloquium on Knowledge Discovery: October 1996
Please note: corrected deadline for submissions
***********************************************
THE INSTITUTION OF ELECTRICAL ENGINEERS
PROFESSIONAL GROUP C4 (ARTIFICIAL INTELLIGENCE)
IN COLLABORATION WITH THE BRITISH COMPUTER SOCIETY
SPECIALIST GROUP ON EXPERT SYSTEMS (SGES)
COLLOQUIUM ON KNOWLEDGE DISCOVERY
LONDON, OCTOBER 17TH-18TH 1996
CALL FOR CONTRIBUTIONS
This colloquium is organised by Professional Group C4
(Artificial Intelligence) of the Institution of Electrical
Engineers, in collaboration with the British Computer Society
Specialist Group on Expert Systems (SGES) and will be held at
the IEE, Savoy Place, London WC2 on October 17th and 18th
1996.
Knowledge Discovery has been defined as 'the non-trivial
extraction of implicit, previously unknown and potentially
useful information from data'. The underlying technologies
include rule induction, case-based reasoning, genetic
algorithms, neural networks and statistics. There is a rapidly
growing body of successful applications of these and other
related technologies in a wide range of areas including
manufacturing, telecommunications, marketing, medicine and
finance.
Contributions are invited on all aspects of Knowledge
Discovery from theoretical issues through to commercial
applications.
Prospective contributors are invited to submit an extended
abstract, outlining the material they propose to present, by
Friday July 12th 1996 at the latest. Speakers will receive
free entry to the colloquium and their travel expenses will be
reimbursed by the IEE.
Abstracts should be sent either by post or by electronic mail
to the colloquium chairman:
Professor Max Bramer, Department of Information Science,
University of Portsmouth, Milton, Southsea PO4 8JF.
Tel: 01705 - 844444 Fax: 01705 - 844006
Email: bramerma@csovax.portsmouth.ac.uk
For all other information, contact:
Ms. Sarah Evans, IEE, Savoy Place, London WC2R 0BL.
Tel: 0171 - 240 - 1871 Fax: 0171-497-3633
Email: sevans@iee.org.uk
KDD Nuggets Index