Knowledge Discovery Nuggets Index


To KDNuggets Directory   |   Here is how to subscribe to KD Nuggets   |   This Year   |   Past Issues

Knowledge Discovery Nuggets(tm) 98:24, e-mailed 98-11-08


News:
  • (text) Won Kim, ACM SIGKDD update
  • (text) Fiona Neil, Quadstone KDD-Cup-98 story
  • (text) Maria Zemankova, NSF KDI: New announcement; deadlines 2/1/99, 5/17/99
  • (text) Amit Seth, Yahoo Data Mining Forum/Club formed

    Publications:
  • (text) Xindong Wu, Re: Incremental Data Mining
  • (text) Mohammed Zaki, Ph.D. Thesis: Scalable Data Mining for Rules

    Tools/Services:
  • (text) Iztok Savnik, Siftware: New version of FDEP
  • (text) James Newton, Request for comments on SEARCH program

    Positions:
  • (text) Ronny Kohavi, San Mateo, CA: Blue Martini Software looking for people
  • (text) Letty Orozco, Austin, Texas: ITC Data Mining Specialist
  • (text) Christophe Giraud, 2 RA Posts in Machine Learning at Bristol, UK
  • (text) D. R. Mani, Mountain View, CA: KDD Job at GTE Govt Systems

    Courses:
  • (text) Bill Goodin, UCLA short course: Data Mining Techniques and
    Applications, January 25-28, 1999, Los Angeles

    Meetings:
  • (text) Belur Dasarathy, SPIE Conference on Data Mining and Knowledge
    Discovery, Orlando, April 1999
  • (text) David Heckerman, Workshop on AI and Statistics,
    Jan 3-6, 1999, Ft. Lauderdale, Florida
    http://uncertainty99.microsoft.com/
    --
    Knowledge Discovery Nuggets (TM) or KDNuggets for short, is an
    electronic newsletter focusing on the latest news, publications, tools,
    meetings, and other relevant items in the Data Mining and Knowledge Discovery
    field. KDNuggets is currently reaching over 5900 readers in 75+ countries
    2-3 times a month.

    Relevant items are welcome and should be emailed to gps
    in ASCII text or HTML format.
    An item should have a subject line which clearly describes
    what is it about to KDNuggets readers.
    Please keep calls for papers and meeting announcements
    short (50 lines or less of up to 80-characters), and provide a web site for
    details, such as papers submission guidelines.
    All items may be edited for size.

    To subscribe, see http://www.kdnuggets.com/subscribe.html

    Back issues of KD Nuggets, a catalog of data mining tools
    ('Siftware'), pointers to data mining companies, relevant websites,
    meetings, etc are available at KDNuggets Directory at
    http://www.kdnuggets.com/

    -- Gregory Piatetsky-Shapiro (editor)
    gps

    ********************* Official disclaimer ***************************
    All opinions expressed herein are those of the contributors and not
    necessarily of their respective employers (or of KD Nuggets)
    *********************************************************************

    ~~~~~~~~~~~~ Quotable Quote ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
    Experience is that marvellous thing that enables you to recognise a mistake
    when you make it again -- F. P. Jones.
    (thanks to http://www.geocities.com/Colosseum/3505/quotations.htm

    Previous  1 Next   Top
    Date: Wednesday, November 04, 1998 11:27 AM
    From: Won Kim [won.kim@cyberdb.com]
    Subject: ACM SIGKDD update
    Web:
  • www.acm.org/sigkdd


    • Here is a brief update on SIGKDD news.

    1. Membership drive -- please join ACM SIGKDD -- see
  • www.acm.org/sigkdd
    • for details.

    2. KDD-99: In addition to KDD-99 main hotel (Marriott Mission Valley),
    ACM signed a contract with an overflow hotel for the KDD-99
    conference -- Double Tree Hotel in Mission Valley, San Diego.
    It is within a one trolley stop from the Marriott Mission Valley.

    The guestroom in each hotel will be offered to KDD-99 attendees
    at $129 a night.

    3. The SIGKDD Awards Committee will consist of
    Jiawei Han (Chair), David Hand, Ross Quinlan, Tom Ditterich, Steve Marron,
    Heikki Mannila, and Hiroshi Motoda.

    The committee will be solciting nominations for
    a) Innovations Award:
    for significant technical innovations in the field
    that have been reduced to practice in significant ways
    or that have significantly influenced direction of research
    and development in the field

    b) Services Award:
    for significant services to the field, that include
    professional volunteer services in disseminating technical
    information to the field, education (i.e., teaching),
    research funding, etc.
    (current ACM SIGKDD officers are not eligible for Services award).

    4. The Editorial Board of the ACM Transactions on Database Systems
    journal has agreed to have TODS publish one 'best TODS-relevant' paper
    nominated by the KDD conference Program Committee each year.


    Previous  2 Next   Top
    Date: Mon, 02 Nov 1998 13:38:34 +0000
    From: Fiona Neil fkn@quadstone.com
    Subject: Quadstone KDD Cup story

    [GPS: Full description of how Quadstone worked on KDD-Cup-98 is available
    at www.kdnuggets.com/meetings/kdd98/quadstone/ ]

    KDD Cup 98:

    Quadstone Take Bronze Miner Award



    Decisionhouse



    Quadstone's Decisionhouse is a complete and integrated suite of software for customer behaviour modelling.
    The Decisionhouse software incorporates all the necessary elements of
    database connectivity, analysis, statistics, visualisation, and data mining for
    business-focused customer modelling. Decisionhouse is used to
    understand customer behaviour for applications including:



  • For more information about Decisionhouse visit the Quadstone web pages at www.quadstone.com




    • Task




    The dataset for this year's competition was provided by the Paralysed Veterans of America (PVA).
    The PVA provides programs and services for US veterans and is one of the largest
    direct mail fund raisers in the US.




    The competition dataset consisted of 191,779 lapsed donor records who received a mailing as part of
    larger campaign sent to a total of 3.5 million donors. Lapsed donors were an important group as the longer
    someone goes without donating the less likely they are to donate again. It was therefore vital for the PVA
    to reactivate these donors.




    Competitors were tasked with developing a model that would help the PVA maximize the net
    revenue generated from future renewal mailings to these lapsed donors. Competitors were assessed on
    the predicted net donations from the campaign.



    Full story of Quadstone methodology is at
    at www.kdnuggets.com/meetings/kdd98/quadstone/

    Previous  3 Next   Top
    Date: Wed, 4 Nov 1998 09:24:39 -0500
    From: Maria Zemankova mzemanko@nsf.gov
    Subject: NSF KDI: New announcement; deadlines 2/1/99, 5/17/99

    NATIONAL SCIENCE FOUNDATION

    KDI: Knowledge and Distributed Intelligence in the Information Age
    Proposal Solicitation, NSF 99-29 (replaces NSF 98-55)
    http://www.nsf.gov/cgi-bin/getpub?nsf9929

    DEADLINES:

    PREPROPOSALS (REQUIRED): FEBRUARY 1, 1999
    FULL PROPOSALS: MAY 17, 1999

    Anticipated date of award: September 1999

    INTRODUCTION:

    Recent advances in computer power and connectivity are reshaping
    relationships among people and organizations, and transforming the
    processes of discovery, learning, and communication.These advances create
    unprecedented opportunities for providing rapid and efficient access to
    enormous amounts of knowledge and information; for studying vastly more
    complex systems than was hitherto possible; and, for increasing in
    fundamental ways our understanding of learning and intelligence in living
    and engineered systems.NSF's Knowledge and Distributed Intelligence (KDI)
    theme is a Foundation-wide effort to promote the realization of these
    opportunities.Proposals are solicited from individuals or groups for
    research that is inherently multidisciplinary or that, while lying within a
    single discipline, has clear impact on at least one other discipline.With a
    budget of approximately $50 million, KDI anticipates funding 40-50
    proposals of varying size and duration.

    To achieve the aims of KDI, proposals are solicited from individuals or
    groups for research that is inherently multidisciplinary or that, while
    lying within a single discipline, has clear impact on at least one other
    discipline. (Throughout, the term multidisciplinary is intended to include
    interdisciplinary and cross-disciplinary research.)

    In FY 1999, KDI will have three foci:
    * Knowledge Networking (KN);
    * Learning and Intelligent Systems (LIS); and
    * New Computational Challenges (NCC).

    This document describes the three KDI foci, and serves as a solicitation
    for proposals in all three areas. We anticipate that research on many
    important problems will span the foci of KN, LIS, and NCC.


    CONTACTS FOR ADDITIONAL INFORMATION

    Further information on KDI can be found at www.nsf.gov/kdi.

    General inquiries should be made via email to kdi@nsf.gov.

    For questions related to use of FastLane, contact FastLane Project Officer,
    703-306-1145, e-mail: fastlane@nsf.gov.

    Cognizant Program Officer(s): Dr. Richard Hilderbrandt, Program Officer,
    Room 1055S, MPS/CHE, telephone 703-306-1844, e-mail: rhilderb@nsf.gov.



    Previous  4 Next   Top
    Date: Sun, 1 Nov 1998 13:17:01 EST
    From: Amit Seth, AmitSeth@aol.com
    Subject: Yahoo Data Mining Forum/Club formed

    Where do you go if you have a question on Data Mining? A lot of newsgroups
    are present today that handle a variety of areas, but none devoted to this
    topic.
    The forum called 'Data Mining Club' would help by being a public place
    to exchange ideas and/or ask questions on the topic of Data Mining.

    You can take a look at the club by going to:
    http://clubs.yahoo.com/clubs/datamining

    Please register for the club and also please help answer some questions
    asked here.

    Previous  5 Next   Top
    Date: Sat, 31 Oct 1998 10:29:27 -0700 (MST)
    From: xwu@gauss.Mines.EDU (Xindong Wu)
    Subject: Re: Incremental Data Mining

    Further to Foster Provost's posting in KDNuggets 98:23 on incremental
    data mining, I have some general comments and a few references to add.

    Incremental data mining is not the same as incremental learning. The
    latter generally handles training data items one by one, while the
    former handles subsets (or samples) of a database one by one
    sequentially. Quinlan's windowing technique (as mentioned in Foster
    Provost's posting) is along this line. There have been discussions
    about the costs vs benefits of windowing, and [Wirth and Catlett 1998]
    has been cited in many places in this regard.

    - J. Wirth and J. Catlett, Experiments on the Costs and Benefits of
    Windowing in ID3, Proceedings of the Fifth International Conference
    on Machine Learning, J. Laird (Ed.), Morgan Kauffman, 1988.

    Incremental batch learning [Clearwater et al 1989] and multi-layer
    incremental induction [Wu & Lu 1998] construct classifications through
    data sampling/partitioning and knowledge refinement across more than
    one learning run. Some research efforts have concentrated on the
    run-time complexity (as mentioned in Foster Provost's posting), and
    others aim to achieve better predictive accuracy in noisy domains,
    because data partitioning is expected to dilute the effects of noise
    into data subsets.

    - S.H. Clearwater, T.P. Cheng, H. Hirsh, and B.G. Buchanan,
    Incremental Batch Learning, Proceedings of the Sixth International
    Workshop on Machine Learning, Morgan Kaufmann, 1989.
    - X. Wu and W. Lo, Multi-Layer Incremental Induction, Proceedings of
    the 5th Pacific Rim International Conference on Artificial
    Intelligence, Springer-Verlag, 1998.

    Recent research efforts on learning ``ensembles of classifiers'' are
    relevant to but significantly different from incremental data mining.
    Learning ``ensembles of classifiers'' such as bagging and boosting
    generates multiple versions of a predictor by running an existing
    learning algorithm many times on a set of re-sampled data, and
    combines them to get an aggregated prediction. Each version of the
    predictor is generated from a sample of the original database, and a
    data item in the original database can be used in many samples for
    generating different versions of the predictor. In incremental data
    mining (at least in multi-layer incremental induction), each data item
    in the original database is partitioned into only one subset, and used
    only once in the induction process.

    Similar to bagging and boosting, stacked generalization [Wolpert 1992,
    Ting & Witten 1997] uses a high-level model to combine lower-level
    models to achieve greater predictive accuracy. These lower-level
    models are generated also by sampling the original database, and
    cross-validation is used to generate higher-level data for the
    high-level model.

    - D.H. Wolpert, Stacked Generalization, Neural Networks, 5 (1992),
    241--259.
    - K.M. Ting & I.H. Witten, Stacked Generalization: When Does It Work?
    Proceedings of the 15th International Joint Conference on
    Artificial Intelligence (IJCAI-97), 866--871.


    Previous  6 Next   Top
    Date: Fri, 30 Oct 1998 19:28:35 -0500 (EST)
    From: Mohammed Zaki zaki@cs.rpi.edu
    Subject: Ph.D. Thesis: Scalable Data Mining for Rules
    Web: http://www.cs.rpi.edu/~zaki.

    My thesis is now available on-line, at http://www.cs.rpi.edu/~zaki.
    It describes scalable and parallel solutions for association rules,
    sequence mining and decision tree classification. Abstract follows.

    - Mohammed Zaki
    Asst. Prof., CS Dept., Rensselaer Polytechnic Inst., Troy NY
    ===================================================================
    Scalable Data Mining for Rules
    Mohammed J. Zaki
    Ph.D. Thesis, University of Rochester, July 1998

    Data Mining is the process of automatic extraction of novel, useful,
    and understandable patterns in very large databases. High-performance
    scalable and parallel computing is crucial for ensuring system
    scalability and interactivity as datasets grow inexorably in size and
    complexity. This thesis deals with both the algorithmic and systems
    aspects of scalable and parallel data mining algorithms applied to
    massive databases. The algorithmic aspects focus on the design of
    efficient, scalable, disk-based parallel algorithms for three key rule
    discovery techniques --- Association Rules, Sequence Discovery, and
    Decision Tree Classification. The systems aspects deal with the
    scalable implementation of these methods on both sequential machines
    and popular parallel hardware ranging from shared-memory systems (SMP)
    to hybrid hierarchical clusters of networked SMP workstations.

    The association and sequence mining algorithms use lattice-theoretic
    combinatorial properties to decompose the original problem into small
    independent sub-problems that can be solved in main memory. Using
    efficient search techniques and simple intersection operations all
    frequent patterns are enumerated in a few database scans. The parallel
    algorithms are asynchronous, requiring no communication or
    synchronization after an initial set-up phase. Furthermore, the
    algorithms are based on a hierarchical parallelization, utilizing both
    shared-memory and message-passing primitives. In classification rule
    mining, we present disk-based parallel algorithms on shared-memory
    multiprocessors, the first such study. Extensive experiments have
    been conducted for all three problems, showing immense improvement
    over previous approaches, with linear scalability in database size.

    Previous  7 Next   Top
    Date: Mon, 2 Nov 1998 09:09:22 +0100 (MET)
    From: Iztok Savnik, Iztok.Savnik@fri.uni-lj.si
    Subject: Siftware: New version of FDEP
    Web: http://infolab.fri.uni-lj.si/~savnik/fdep.html

    We would like to announce thru the KDD-Nuggets distribution
    list a new version of FDEP - program for the discovery of
    functional dependencies from relations. In comparison to
    the previous version (v1.0), FDEP 1.1 includes the following
    improvements:

    - novel algorithm for the discovery of functional dependencies,
    - support for unknown values,
    - method for handling noisy data,
    - sample domains are included in the distribution,
    - increased maximal number of relation tuples, and,
    - increased maximal number of relation attributes.

    FDEP distribution (including source code, documentation
    and sample relations) can be obtained at
    http://infolab.fri.uni-lj.si/~savnik/fdep.html

    Previous  8 Next   Top
    Date: Mon, 2 Nov 1998 09:54:49 -0500 (EST)
    From: james newton morgan jnmorgan@umich.edu
    Subject: Request for comments on some proposed enhancements to SEARCH program

    The SEARCH program from OSIRIS (formerly Automatic Interaction
    Detector or AID) is available in MicrOsiris for the PC from
    vaneckn@philacol.edu. We are thinking of some enhancements to it and
    would appreciate comments. It is designed for analyzing surveys or
    subsamples with 1,000-30,000 cases where there is a criterion variable
    or classification to be accounted for. It also allows a covariance
    search where there is one dominant explanatory factor like income,
    gender or race. We propose the following enhancements:

    1. In addition to a tree diagram, a hierarchical summary table with
    one line for each group, indented for subgroups and with final groups
    indicated. This would allow more information about each group (cases,
    mean, distribution for a categorical criterion, or simple regression
    for covariance analysis). It should be simple for the user to alter
    the table to make the definitions of each split more understandable.

    2. For a ranked dependent 'variable', a split criterion based on rank
    correlation, probably Kendall's Tau-b. Using the likelihood ratio chi
    square designed for a categorical criterion loses information about
    rank order.

    3. Extend the facility to produce the recode for expected values or
    residuals to include categorical dependent variables (chi or rank),
    where the expected values would be a distribution rather than a mean.

    4. Make the use of a 'runfile' easier, with the whole analysis
    strategy including data source, filter, recode, and analysis, in a
    separate file so there is a record and one can alter for reruns and
    save for the files.

    5. Provide a fourth possible stopping rule based on a miniumu t-ratio
    or maximum null probability, even though such tests are improper after
    extensive searching and with survey data with design effects. (And
    exclusive reliance on significance can isolate very small homogenous
    groups which would be useless in predicting back to the population.)
    Currently one can specify minimum final group size, maximum number of
    splits, and/or minimum reduction in unexplained variance relative to
    original total.

    6. Since datafiles are reasonably standard rectangular asci files, but
    dictionaries are not, stay with separate data and dictionary files,
    but provide the recode to convert SAS or SPSS and perhaps some other
    dictionaries to an OSIRIS dictionary.

    james n. morgan e-mail: jnmorgan@umich.edu
    institute for social research phone 313 764 8388
    po box 1248 home 313 668 8304
    ann arbor, mi 48109 fax 313 747 4575

    Previous  9 Next   Top
    Date: Mon, 02 Nov 1998 13:22:38 -0800
    From: Ronny Kohavi ronnyk@bluemartini.com
    Subject: Ron Kohavi at Blue Martini Software looking for people
    Web: http://www.bluemartini.com/

    Blue Martini Software, a startup combining two of the hottest
    technologies: data mining and e-commerce, is hiring.

    Ronny Kohavi, director of data mining applications at Blue Martini,
    is looking for at least two engineers to join the company and
    work on the data mining component of the system.

    Responsibilities include developing, modifying, interfacing machine
    learning and statistical algorithms, and integrating them into the larger
    e-commerce solution.

    Applicants should have extensive programming experience with C++, an
    MS or Ph.D. degree in Computer Science (or equivalent), and prior knowledge
    of machine learning/data mining.

    Interested? Send e-mail to ronnyk@bluemartini.com and CC hr@bluemartini.com

    More information:

    - Before joining Blue Martini Software, Ronny kohavi managed the
    MineSet(TM) data mining and visualization project at Silicon Graphics
    http://mineset.sgi.com, using MLC++ http://www.sgi.com/Technology/mlc/,
    the Machine Learning library in C++, which he headed at Stanford University
    and then at SGI.
    - Information about Blue Martini can be found at http://www.bluemartini.com/
    Look at the board of directors and management and you'll be impressed.
    We are located in San Mateo, California.

    -- Ronny Kohavi

    Previous  10 Next   Top
    Date: Thu, 5 Nov 1998 12:01:39 -0600
    From: Letty Orozco letty@itctx.com
    Subject: Job: TX-US-Austin Data Mining Specialist

    Intelligent Technologies Corporation (ITC) provides fraud detection
    solutions for healthcare and insurance industries. ITC has developed
    end-to-end solutions for Medicaid fraud management and tax compliance
    management based on advanced pattern recognition technologies. ITC's
    products and related services have saved businesses and consumers hundreds
    of millions of dollars annually by detecting and helping prevent fraudulent
    transactions. ITC is looking for key individuals to become part of a highly
    motivated team working with hot technologies in a company with great growth
    potential. Be a part of the success.

    DATA MINING SPECIALIST

    The Austin, Texas office of ITC is looking for a candidate with applied
    mathematics/engineering/CS degree and 2-5 years of experience in data
    mining. Candidates not requiring relocation to Texas are preferred. The
    ideal candidate will have a broad background and the following skill sets.

    * Applications of neural networks, decision trees, fuzzy logic and statistics
    * Experience in commercial data mining tools (Clementine, SAS, etc.)
    * Working knowledge of relational data bases and SQL
    * Good communication skills

    We offer excellent compensation plus a complete benefits package including
    an employee stock option plan. For consideration, send your resume
    including salary history to: ITC, Job Code E, 9015 Mountain Ridge Drive,
    Suite 350, Austin, TX, 78759. Or fax to (512) 343-1608. Or e-mail to:
    hello@itctx.com. Check out our Web site at: http://www.itctx.com

    Previous  11 Next   Top
    Date: Wed, 21 Oct 1998 13:34:43 +0100 ()
    From: Christophe Giraud-Carrier cgc@cs.bris.ac.uk
    Subject: 2 RA Posts in Machine Learning at Bristol

    The Machine Learning Research Group at the University of Bristol
    offers two Research Assistantships. One assistantship (RA1) is for a
    36-month duration and must be filled as soon as possible. The other
    (RA2) is for a 26-month duration and would start by 1 May 1999.

    The 2 assistantships focus on machine learning and data mining, and
    are part of a European Long-term Reactive Research Project entitled
    'METAL: A Meta-Learning Assistant for Providing User Guidance in
    Machine Learning and Data Mining'. The main research themes of the
    project are: model selection (choice of an appropriate learning
    model/algorithm for a given application task), data screening and
    data transformation prior to knowledge extraction, meta-learning
    approaches to model selection and data transformation. The project
    involves 6 partners (2 industrial and 4 academic) and its results
    will be implemented in the commercial system Clementine.

    Ideally, candidates should have: (1) a Ph.D. in Computer Science;
    (2) a strong background in artificial intelligence, machine learning
    (both symbolic and numeric), and data modelling/analysis/mining;
    and (3) experience in software development (e.g., implementing
    prototypes).

    Starting salary for both posts will be on the Grade 1A scale, up to
    point 5 depending on qualifications (between 15,462 and 18,864 sterling
    pounds per annum). If you wish to be considered for any of these
    positions, please send your curriculum vitae, together with the names
    and addresses of 2 references, to the following contact person:

    Dr. Christophe Giraud-Carrier cgc@cs.bris.ac.uk

    Electronic applications are encouraged for reasons of speed; please
    specify the reference of the post for which you are applying.

    For those with no e-mail facilities, applications can be sent by
    fax or regular mail to the above contact person (address and
    fax number below):

    Department of Computer Science
    University of Bristol
    Merchant Venturers Building
    Woodland Rd
    Bristol, BS8 1UB
    United Kingdom
    Fax: +44-117-954-5208

    Applications will be accepted until the positions are filled.

    Previous  12 Next   Top
    Date: Tue, 03 Nov 1998 15:26:16 -0500
    From: 'D. R. Mani' mani@gte.com
    Subject: KDD Job Opening at GTE Govt Systems, Mountain View, CA

    An Applied Researcher/Developer needed to lead the
    Knowledge Discovery initiative at GTE Government Systems

    RESPONSIBILITIES: Lead the Knowledge Discovery initiative at GTE
    Government Systems in the Electronic Systems Division (ESD).
    Participate in the design and development of a wide range of
    state-of-the-art systems for data mining and knowledge discovery --
    including systems for information assurance, intrusion detection and
    security, electronic commerce and multi-media data mining. Develop
    and grow in-house knowledge discovery expertise through additional
    training and hiring.

    QUALIFICATIONS: The ideal candidate will have a Master's or Ph.D
    Degree in Computer Science or a related field with proven technical
    abilities in the areas of knowledge discovery and datamining (KDD),
    machine learning (neural networks, decision trees, genetic
    algorithms, clustering, etc.) and artificial intelligence
    (rule-based systems, expert systems, etc.). In addition to
    technical excellence, specific experience in dealing with complex
    real-world problems and large data sets is important. Good
    communication and presentation skills for technical, semi-technical
    and non-technical audiences is a strong asset.

    United States citizenship is required in order to obtain the highest
    levels of security clearance.

    WORK ENVIRONMENT: The Electronics Systems Division (ESD) at GTE
    Government Systems engineers specialized, state-of-the-art
    telecommunications and data communication systems, and develops
    knowledge management and assurance infrastructure for these systems
    over intranet, extranet and Internet environments. ESD subscribes to
    a customer-centric business model, and uses technical expertise,
    talent and innovation to support customer needs. Projects involve
    use of a wide range of cutting-edge technologies, contributing to a
    challenging (and sometimes demanding), team-oriented and proactive
    work environment.

    ESD promotes a professional, task oriented environment with the
    option for flexible hours/telecommuting and excellent benefits
    (including every other Friday off). ESD is head quartered in
    Mountain View CA, with locations in Washington DC, Baltimore MD,
    Thousand Oaks CA, and Colorado Springs CO.

    CONTACT INFORMATION:
    Please send a resume and a cover letter to
    Wade F. Schott
    Director, Technology and Product Development
    100 Ferguson Drive
    P. O. Box 7188 MS 7316
    Mountain View, CA 94039

    Electronic submissions should be directed to:
    wade.schott@gsc.gte.com

    Previous  13 Next   Top
    Date: Fri, 30 Oct 1998 17:41:23 -0800
    From: Bill Goodin, bgoodin@unex.ucla.edu
    Subject: UCLA short course on 'Data Mining Techniques and Applications'
    Web: http://www.unex.ucla.edu/shortcourses

    On January 25-28, 1999, UCLA Extension will present the short course,
    'Data Mining Techniques and Applications' on the UCLA campus in
    Los Angeles.

    The instructors are Wei-Min Shen, PhD, USC Information Sciences
    Institute; Rakesh Agrawal, PhD, IBM Almaden Research Center; and
    Jiawei Han, PhD, Simon Fraser University.

    This course is intended for scientists, engineers, and information
    managers who need to learn and apply data mining techniques (tools
    for discovering valuable knowledge from very large data sets) to their

    scientific research, system design, business management, or any
    other related applications. The lecturers are among the world-leading
    experts in the field with extensive experience in basic research as
    well as in real industrial application.

    This course should enable participants to understand and have hands-on
    experience in:
    o Basic concepts of data mining
    o The overall process of data mining
    o Critical steps in the data mining process
    o Relationships between data mining and other scientific disciplines
    o Formalizing data mining problems
    o Data preprocessing
    o Data classification
    o Data clustering
    o Database structures and their operations
    o Time serial data analysis
    o Visualization
    o Prediction and forecasting

    The course fee is $1395, which includes extensive course materials.
    These materials are for participants only, and are not for sale.

    For additional information and a complete course description, please
    contact Marcus Hennessy at:

    (310) 825-1047
    (310) 206-2815 fax
    mhenness@unex.ucla.edu
    http://www.unex.ucla.edu/shortcourses

    This course may also be presented on-site at company locations.

    Previous  14 Next   Top
    Date: Fri, 30 Oct 1998 17:24:21 -0800 (PST)
    From: Dr. Belur V. Dasarathy belur@yahoo.com
    Subject: SPIE Conference on Data Mining and Knowledge Discovery, Orlando, April 1999
    Web: http://members.tripod.com/~belur/kddprog.html

    The Final program, including dates/times etc., for the SPIE Conference
    on Data Mining and Knowledge Discovery: Theory, Tools, and Technology,
    being held in Orlando, in April 1999 has now been published on the web
    at http://members.tripod.com/~belur/kdd.html

    Click on 'Final Program' to see the details.
    Belur V. Dasarathy
    Conference Chairman

    Previous  15 Next   Top
    Date: Fri, 30 Oct 1998 13:24:35 -0800
    From: David Heckerman heckerma@MICROSOFT.com
    Subject: Call for participation: Workshop on AI and Statistics, Jan 1999, Florida
    Web: http://uncertainty99.microsoft.com/

    Call For Participation
    Uncertainty 99: Seventh International Workshop on
    Artificial Intelligence and Statistics

    January 3-6, 1999
    Ft. Lauderdale, Florida

    http://uncertainty99.microsoft.com/

    PURPOSE:
    This is the seventh in a series of workshops that has brought together
    researchers in Artificial Intelligence and in Statistics to discuss
    problems of mutual interest. The exchange has broadened research in
    both fields and has strongly encouraged interdisciplinary work.

    Attendance at the workshop is *not* limited to paper presenters.

    TECHNICAL PROGRAM:
    To encourage interaction and a broad exchange of ideas, there will be
    20 discussion papers in single session meetings over three days
    (Jan. 4-6). A poster session and brief plenary summaries of
    posters prior to these sessions will provide the means for presenting
    and discussing the remaining research papers. Papers accepted for
    plenary and poster presentation are listed at the web site listed above.

    [...edited. GPS]

    For more information, including
    Registration forms, a provisional program, a list of the program
    committee, and other details about the conference and the Society for
    Artificial Intelligence and Statistics can be found at
    http://uncertainty99.microsoft.com/.

    David Heckerman and Joe Whittaker,
    Conference Chairs
    Previous  16 Next   Top