home *** CD-ROM | disk | FTP | other *** search
/ OS/2 Shareware BBS: 16 Announce / 16-Announce.zip / 294249.zip / 294249.ANN
Text File  |  1994-05-03  |  26KB  |  405 lines
  1. DATE       940503
  2. CATEGORY   GD00, GD60, G120, GD70, WS00, WS20, WS25, WS50
  3. TITLE      IBM NEURAL NETWORK UTILITY FOR OS/2 AND WINDOWS
  4. ABSTRACT          Neural networks and fuzzy logic are powerful new application
  5.            development technologies.  Neural networks can find patterns and
  6.            learn relationships in business data.  They differ from traditional
  7.            programming in that the neural network training process is automated,
  8.            providing a cost-effective way of turning data into useful processing
  9.            functions.  Fuzzy logic is an exciting extension of the well-known,
  10.            rule-based expert system methodology, with the advantage of letting
  11.            experts represent their knowledge in their own language.  The Neural
  12.            Network Utility (NNU) product family brings the power and
  13.            productivity of these advanced technologies to commercial application
  14.            developers working in OS/2 (R) and Windows (1) computing
  15.            environments.
  16.                   Neural Network Utility for OS/2 (NNU/2) and Neural Network
  17.            Utility for Windows (NNU/Win) Version 3.1 are enhanced and repackaged
  18.            members of the Neural Network Utility product family.   The NNU/2 and
  19.            NNU/Win products are functionally equivalent.  They can be used as
  20.            stand-alone development tools or as part of a client/server
  21.            environment.  Enhancements include support for additional neural
  22.            network models, for fuzzy rule processing, and for client/server
  23.            development and remote data access from AS/400 (R) system and RISC
  24.            System/6000 (R) servers.  Increased function and ease-of-use in the
  25.            data translation and control script language yield improved
  26.            productivity.
  27.                   Applications developed using NNU/2 and NNU/Win are portable
  28.            between all NNU environments.  NNU provides a language-independent
  29.            API for embedding neural network and fuzzy rule processing into OS/2
  30.            and Windows application programs.
  31.                   Refer to the following Software Announcements, dated
  32.            May 3, 1994, for additional information.
  33.            o   294-248 Neural Network Utility/400
  34.            o   294-251 AIX Neural Network Utility/6000
  35.            o   294-250 Neural Network Utility Entry for OS/2 and Windows
  36.                   One-Time Charge:  $1,500
  37.                   Planned Availability Date:  September 9, 1994
  38.             (R) Registered trademark of International Business Machines
  39.               Corporation
  40.             (1) Trademark of Microsoft Corporation
  41.            IN BRIEF . . .
  42.                   Features
  43.            o   Neural network application development support
  44.            o   Client/Server product orientation
  45.            o   Knowledge-based processing using fuzzy rules
  46.            o   Enhanced application developer productivity tools
  47.                   Neural Network Utility Family
  48.            o   Neural Network Utility for OS/2 (NNU/2)
  49.            o   Neural Network Utility for Windows (NNU/Win)
  50.            o   Neural Network Utility/400 (NNU/400)
  51.            o   AIX (R) Neural Network Utility/6000 (NNU/6000)
  52.            o   Neural Network Utility Entry for OS/2 (NNU Entry/2)
  53.            o   Neural Network Utility Entry for Windows (NNU Entry/Win)
  54. PRODNO       5871-AAA  03H3892
  55.              5871-AAA  03H3893
  56. OVERVIEW   DESCRIPTION
  57.            NEURAL NETWORK APPLICATION DEVELOPMENT SUPPORT
  58.            NNU for OS/2 and NNU for Windows allow users to develop neural
  59.            network applications on PCs and deploy them either as stand-alone PC
  60.            applications or as multi-user AS/400 system or RISC System/6000
  61.            solutions when used with NNU/400 and NNU/6000.
  62.            ENABLE CUSTOMER'S COMPETITIVE ADVANTAGE:  Neural network technology
  63.            detects patterns and relationships in data and then uses that
  64.            information to perform data processing functions.  In cases where the
  65.            relationships are complex or unknown, neural networks open new areas
  66.            to application development.  Leading-edge applications using this
  67.            technology have provided significant technical advantages in the
  68.            retail, marketing, finance, manufacturing, and health industries.
  69.            APPLICATION DEVELOPER PRODUCTIVITY:  Application development time for
  70.            neural network applications can be significantly less than for
  71.            traditional data processing applications.  A typical neural network
  72.            application is developed by doing problem analysis, data analysis and
  73.            scaling, training the neural network, and then testing and
  74.            validation.  There are no algorithm design or code and debug steps in
  75.            the neural network application development process, resulting in
  76.            shorter development times.
  77.            CLIENT/SERVER PRODUCT ORIENTATION
  78.            The client/server function in NNU/2 and NNU/Win allows a user to
  79.            develop applications on a PC while the neural network processing is
  80.            performed on an AS/400 system or RISC System/6000 server.  It also
  81.            allows remote data access.
  82.            ACCESS TO ENTERPRISE DATA:  Using NNU in client/server mode, an
  83.            application developer can access databases and files on the server
  84.            machines running NNU/400 or NNU/6000.  This reduces the duplication
  85.            of data and the workstation storage requirements, along with data
  86.            integrity and security problems.
  87.            INCREASED CAPACITY/PERFORMANCE IMPROVEMENT:  The client/server mode
  88.            allows a developer running OS/2 or Windows on a PC to run the neural
  89.            network computations on a larger AS/400 system or RISC System/6000
  90.            machine.  For the computationally demanding training process, this
  91.            mode of operation allows more and larger neural networks to be
  92.            developed in the same amount of time.  As a result, users can solve
  93.            more difficult problems involving more variables.
  94.            KNOWLEDGE-BASED PROCESSING USING FUZZY RULES
  95.            Fuzzy rule processing adds a knowledge-based processing approach to
  96.            complement the data-based neural network approach.
  97.            ENABLE CUSTOMER'S COMPETITIVE ADVANTAGE:  Rule-based expert systems
  98.            have been widely used in business applications.  The two areas of
  99.            weakness are the difficulty maintaining large rule bases and the
  100.            sometimes slow execution times.  Fuzzy rule systems typically allow a
  101.            problem to be solved using fewer rules, and the execution times are
  102.            fast because search algorithms are not used.  Fuzzy rules are
  103.            translated into mathematical form and solved by the fuzzy inference
  104.            engine.  This technology allows domain experts to represent their
  105.            knowledge using their own terminology and quickly develop
  106.            knowledge-based applications.
  107.            ENABLING NEW APPLICATIONS:  The addition of fuzzy rule processing to
  108.            the existing neural network capabilities of NNU creates the
  109.            possibility of new applications using the pattern recognition and
  110.            modeling capabilities of neural networks along with the rule
  111.            processing of fuzzy rule systems.  The fuzzy rule editor in NNU
  112.            allows a developer to quickly define fuzzy variables, fuzzy sets and
  113.            membership functions, and corresponding fuzzy rules.  This
  114.            easy-to-use environment leads to rapid prototyping and evaluation of
  115.            the expert-system application.
  116.            ENHANCED APPLICATION DEVELOPER PRODUCTIVITY TOOLS
  117.            The NNU built-in data translation tools and scripting language for
  118.            automated training have been enhanced to provide improved ease-of-use
  119.            and increased function.
  120.            APPLICATION DEVELOPER PRODUCTIVITY:  The NNU translate editor and
  121.            translate filter provide a powerful template-based function for
  122.            performing data translation and scaling of neural network training
  123.            data.  A large percentage of the neural network application
  124.            development cycle is spent working with data.  NNU splits source
  125.            files into training and testing data automatically, generates
  126.            translation templates by scanning the source data, provides more than
  127.            20 mathematical functions, including vector operations, and provides
  128.            an easy-to-use editor for specifying the translation operations.
  129.            This allows developers to quickly and easily create what-if scenarios
  130.            to explore data representation issues.
  131.                   The NNU control script editor and language allow a developer
  132.            to specify an automated sequence of actions for training and testing
  133.            neural network applications.  Neural network parameters can be set,
  134.            data sources can be selected, network performance can be logged, and
  135.            breakpoints can be set.  New features include text script files that
  136.            can be edited in any text editor, variable support so that re-usable
  137.            scripts can be written, and looping control constructs.  The NNU
  138.            control script language allows automated development of neural
  139.            network applications.
  140.            PRODUCT POSITIONING
  141.            The NNU family of products is a complete set of development tools
  142.            that combine neural network and fuzzy rule system technologies that
  143.            can be used to create leading-edge business solutions.  The NNU
  144.            products run on a wide range of platforms from Windows and OS/2 on
  145.            PCs to AIX on RISC System/6000 workstations, up to AS/400 mid-range
  146.            systems.  This wide coverage, along with client/server capabilities,
  147.            allows a developer to choose the most appropriate platform for
  148.            development and deployment.
  149.                   Workstation-based development environments use graphical user
  150.            interfaces to allow point-and-click construction of applications.
  151.            NNU provides tools for local and remote database access, data
  152.            translation and scaling, visualization of neural networks and fuzzy
  153.            sets, and a scripting language for automated training sequences.  It
  154.            supports multiple neural network models, and provides programming
  155.            interfaces for user-defined custom models.
  156.                   Neural Network Utility can be used across multiple industries
  157.            and enables the development of a wide range of applications.  NNU
  158.            provides a cost-effective way of turning data into meaningful
  159.            processing functions using neural networks and turning knowledge into
  160.            application logic using fuzzy rule systems.  Applications range from
  161.            risk assessment and portfolio management in finance to fraud
  162.            detection and policy assessments in health and insurance, all the way
  163.            to optimization, scheduling, and visual inspection systems in
  164.            manufacturing and process control.
  165.                   The Neural Network Utility products should be used by
  166.            application development professionals with neural network and fuzzy
  167.            rule system education or experience who want to develop robust
  168.            commercial applications using these technologies.
  169. CUSTINFO   PUBLICATIONS
  170.                   The NEURAL NETWORK UTILITY INSTALLATION AND USER'S GUIDE
  171.            (SC41-0223) is shipped with the product.
  172.                   Additional copies of unlicensed publications will be available
  173.            for a fee after product availability.  These copies may be ordered
  174.            from your IBM representative, through the system library subscription
  175.            service (SLSS), or by direct order.
  176.                                                         ORDER
  177.            TITLE                                        NUMBER
  178.            Neural Network Utility
  179.              Programmers' Reference                     SC41-0222
  180.            EDUCATION SUPPORT
  181.                   The appropriate curriculum will be updated, as necessary, to
  182.            include the enhancements in this announcement.
  183.                   Details of the education support will be provided by product
  184.            availability through separate Education Announcements.
  185.                   Descriptions of all classroom and self-study courses are
  186.            contained in the SKILL DYNAMICS (TM) CATALOG OF IBM EDUCATION
  187.            (G320-1244).
  188.                   Call Skill Dynamics, An IBM Company, at 800-IBM-TEACH
  189.            (426-8322) for schedules and enrollments.
  190.             (TM) Trademark of International Business Machines Corporation
  191. TECHINFO   TECHNICAL INFORMATION
  192.            SPECIFIED OPERATING ENVIRONMENT
  193.            HARDWARE REQUIREMENTS:
  194.            NNU FOR OS/2:
  195.                   Any personal computer supported by IBM OS/2 2.1, or later,
  196.            with:
  197.            o   386DX 33MHz processor (486DX 33MHz recommended)
  198.            o   6MB minimum RAM memory (8MB recommended)
  199.            o   VGA or higher resolution display supported by OS/2.
  200.            o   4MB hard file available free space
  201.            o   For Client/Server mode, any LAN adapter supported by OS/2 2.1, or
  202.                later.
  203.            NNU FOR WINDOWS:
  204.                   Any personal computer supported by Microsoft Windows 3.1, or
  205.            later, compatible version in enhanced mode with:
  206.            o   386DX 33MHz processor (486DX 33MHz recommended)
  207.            o   4MB minimum RAM memory (6MB recommended)
  208.            o   VGA or higher resolution display supported by Windows 3.1.
  209.            o   4MB hard file available free space
  210.            o   For Client/server mode, any LAN adapter supported by Windows 3.1.
  211.            SOFTWARE REQUIREMENTS:
  212.            NNU FOR OS/2:
  213.            o   NNU/2 3.1 will run only under OS/2 2.1, or later.
  214.            o   Custom neural network model and custom interface program
  215.                development requires an OS/2 compiler capable of generating
  216.                32-bit OS/2 2.1 compatible DLLs.
  217.            o   Client/Server development requires IBM TCP/IP for OS/2
  218.                Version 2.0, or later.
  219.            o   NNU/2 local database support requires a DB2/2 (TM) database.
  220.            o   NNU/2 remote database access requires the facilities provided by
  221.                the DB2 (R) family of client/server products.
  222.            NNU FOR WINDOWS:
  223.            o   NNU for Windows 3.1 will run under Microsoft Windows 3.1, or
  224.                later, compatible version.
  225.            o   Custom neural network model and custom interface program
  226.                development requires a compiler capable of generating 32-bit
  227.                Windows compatible DLLs.
  228.            o   Client/Server development requires IBM TCP/IP for DOS
  229.                Version 2.1, or later.
  230.            o   NNU/Win provides generic (local and remote) database support
  231.                through the Open DataBase Connectivity (ODBC) interface.  This
  232.                requires an ODBC driver for each target database.
  233.            COMPATIBILITY:  Neural Network Utility for OS/2 and Neural Network
  234.            Utility for Windows are compatible with other members of the NNU 3.1
  235.            product family.  Neural networks and fuzzy rule systems developed
  236.            using NNU/2 or NNU/Win can be run by other NNU family products.
  237.            PERFORMANCE CONSIDERATIONS:  Neural network application development
  238.            typically requires large amounts of floating-point computations.
  239.            Application developer productivity will be gated by the speed of the
  240.            hardware.  Fuzzy rule system development places no great demands on
  241.            the system.
  242.                   In general, neural network and fuzzy logic applications can be
  243.            deployed on any machine capable of running Windows 3.1 or OS/2 2.1.
  244.            PLANNING INFORMATION
  245.            PACKAGING:  Installing NNU/2 and NNU/Win is easy and can be done in
  246.            less than 30 minutes.  An installation program will guide the user
  247.            through the installation process.
  248.            PACKAGING:  The IBM Neural Network Utility for OS/2 and Neural
  249.            Network Utility for Windows contain the following packaged as a
  250.            single unit:
  251.            PROGRAM PRODUCT
  252.            o   Registration Card
  253.            o   Installation and User Guide (publication)
  254.            o   IBM Program License Agreement on diskette container
  255.            o   Program diskettes (3.5-inch media)
  256.            o   Program License information
  257.            ADDITIONAL LICENSE COPY (WITH CERTIFICATE):  Proof of Additional
  258.            License.
  259.            ADDITIONAL LICENSE COPY (WITHOUT CERTIFICATE):  No material is
  260.            shipped.
  261.            SECURITY, AUDITABILITY AND CONTROL
  262.                   User management is responsible for evaluation, selection and
  263.            implementation of security features, administrative procedures and
  264.            appropriate controls in application systems and communication
  265.            facilities.
  266. ORDERING   ORDERING INFORMATION
  267.                   To order the programs in this announcement for order type,
  268.            specify the order type number, feature number, part number, program
  269.            name, media, and quantity.
  270.            PACKAGES
  271.            ORDER
  272.            TYPE           FEATURE    PART      PROGRAM
  273.            NUMBER         NUMBER     NUMBER    NAME              MEDIA
  274.            5871-AAA       7894       03H3892   NNU
  275.                                                  for OS/2        3.5-inch
  276.            5871-AAA       7895       03H3893   NNU for
  277.                                                  Windows         3.5-inch
  278.            ADDITIONAL LICENSE
  279.            ORDER
  280.            TYPE           FEATURE    PART      PROGRAM
  281.            NUMBER         NUMBER     NUMBER    NAME
  282.            5872-AAA       7898       03H3896   NNU for
  283.                                                  OS/2
  284.                                                  Additional
  285.                                                  License
  286.            5872-AAA       7904       03H3897   NNU for
  287.                                                  Windows
  288.                                                  Additional
  289.                                                  License
  290.            5872-AAA       7907       03H3900   NNU for
  291.                                                  OS/2
  292.                                                  Additional
  293.                                                  License
  294.                                                  Certificate
  295.            5872-AAA       7908       03H3901   NNU for
  296.                                                  Windows
  297.                                                  Additional
  298.                                                  License
  299.                                                  Certificate
  300.            PROGRAM PACKAGE UPGRADES
  301.            ORDER
  302.            TYPE           FEATURE    PART      PROGRAM
  303.            NUMBER         NUMBER     NUMBER    NAME              MEDIA
  304.            5875-UPG       7911       03H3904   NNU for
  305.                                                  OS/2            3.5-inch
  306.            5875-UPG       7912       03H3905   NNU for
  307.                                                  Windows         3.5-inch
  308.            o   Program Package -- program diskette with documentation.
  309.            o   Additional License Copies -- authorization  for the customer who
  310.                has a program package to make a single copy of the program
  311.                diskette and documentation.  There are two ways to receive
  312.                authorization to copy.  Receipt of the invoice authorizes the
  313.                customer to make the Additional License Copy.  The ICA must be in
  314.                effect.  The customer also has the option of ordering a Proof of
  315.                Additional License certificate.  The ICA is not required.
  316.            o   Documentation -- Separate documentation without program
  317.                diskettes.
  318.            o   Program Package Upgrade  -- Program diskette with documentation.
  319.            IBM EDUCATION FULFILLMENT CENTER:  The announced programs are
  320.            eligible to be ordered through the IBM Education Fulfillment Center.
  321.            Please note that availability through the IBM Education Fulfillment
  322.            Center may not coincide with the programming availability dates noted
  323.            in other sections of this document.
  324.                   The IBM Education Fulfillment Center is the order processing
  325.            location for selected Personal Computers and related
  326.            hardware/software for eligible education end-users.
  327.            PROGRAM UPGRADE
  328.                   The program package upgrade option allows currently licensed
  329.            users of Neural Network Utility (5798-RYK) to obtain the new function
  330.            for a program upgrade charge.  The program upgrade to Neural Network
  331.            Utility for OS/2 or Neural Network Utility for Windows will be
  332.            available September 9, 1994.  Orders will be accepted after
  333.            September 9, 1994, if the specific date has not been determined.
  334.            These upgrades will be available through IBM Authorized Personal
  335.            Computer Dealers and Industry Remarketers -- Personal Computers
  336.            certified to market IBM Authorized Advanced Products.  The remarketer
  337.            will require a Proof of License for each upgrade ordered.  The
  338.            original Proof of License is the colored front cover page inside the
  339.            hard binder or the separate Proof of License page in the program
  340.            package, as identified by IBM to the remarketer.
  341.                   Customers may also order upgrades through the local IBM office
  342.            using the IBM Personal Computer/System Program IBM US Upgrade and
  343.            Certification Order Form.
  344.            TERMS AND CONDITIONS
  345.            LICENSING:  The programs in this announcement are licensed under the
  346.            terms and conditions of the IBM Program License Agreement.  A copy of
  347.            the agreement is included in the program package.
  348.            ADDITIONAL LICENSE:  The IBM Customer Agreement (ICA) must be in
  349.            effect, or Proof of Additional License certificate must be acquired.
  350.            LIMITED WARRANTY:
  351.            o   Program:  Yes   Media:  Three months
  352.            o   Additional Licenses:  Yes   Media:  Not applicable
  353.            PROGRAM SERVICES:  Central Service, including the IBM Support Center,
  354.            will be available during the first 90 days following the date of
  355.            delivery to the customer by IBM, but not later than May 31, 1997.
  356.            The statement of service included in the program package describes
  357.            this service.
  358.            VOLUME ORDERS:  Not applicable
  359.                   For Information regarding other volume orders, contact your
  360.            IBM representative.
  361.            TESTING PERIOD:  A 60-day trial period applies for up to three
  362.            licenses.
  363.                   The ICA must be in effect.
  364.            NATIONAL EDUCATION LICENSE FEE:  A National Education License Fee
  365.            (NELF) is available under the terms and conditions of the Education
  366.            Allowance Attachment.  The NELF, which represents a 20% discount off
  367.            the single unit price, is not subject to any further discount or
  368.            allowance.
  369.            CHARGES
  370.            PROGRAM        FEATURE                                      ONE-TIME
  371.            NUMBER         NUMBER    ORDER TYPE                         CHARGE
  372.            03H3892        7894      NNU for OS/2
  373.                                       Program Package                  $1,500
  374.            03H3893        7895      NNU for Windows
  375.                                       Program Package                   1,500
  376.            03H3896        7898      NNU for OS/2
  377.                                       Additional License                1,320
  378.            03H3897        7904      NNU for Windows
  379.                                       Additional License                1,320
  380.            03H3900        7907      NNU for OS/2
  381.                                       Additional License
  382.                                       Certificate                       1,320
  383.            03H3901        7908      NNU for Windows
  384.                                       Additional License
  385.                                       Certificate                       1,320
  386.                                                                        ONE-TIME
  387.                                                                        LICENSE
  388.                                                                        CHARGE
  389.            PART         FEATURE   UPGRADE               UPGRADE        FOR
  390.            NUMBER       NUMBER    TO                    FROM           UPGRADE
  391.            03H3904      7911      NNU for
  392.                                     OS/2                NNU/2            $150
  393.            03H3905      7912      NNU for
  394.                                     Windows             NNU/2             150
  395.            ADDITIONAL LICENSES CHARGE:  Customers who pay a one-time charge for
  396.            additional licenses may copy machine-readable program materials and
  397.            printed documentation previously acquired from IBM in a program
  398.            package.  One copy may be made for each additional license ordered.
  399.            PROGRAM UPGRADE CHARGE:  A program upgrade charge will apply for
  400.            upgrades from Neural Network Utility (5798-RYK) to Neural Network
  401.            Utility for OS/2 or Neural Network Utility for Windows.
  402.                   The charges are unchanged by this announcement.
  403.  
  404.               - - -    E N D   O F   P R I N T O U T    - - -
  405.