Loading ...
Sorry, an error occurred while loading the content.

EasyNN-plus V7.0a beta 1

Expand Messages
  • steve@easynn.com
    The first beta release of EasyNN-plus 7.0a is now available for testing. Most changes are related to performance. Previous performance improvements in versions
    Message 1 of 2 , Oct 1, 2005
    • 0 Attachment
      The first beta release of EasyNN-plus 7.0a is now available for
      testing.

      Most changes are related to performance.

      Previous performance improvements in versions 2 to 6 have been
      consolidated into common methods. The best example of this is that all
      progressive array indexing has been replaced by link lists. Almost all
      fixed arrays are replaced by extendable double link lists.

      Learning performance has been improved by propagating the square error
      rather that the absolute error. This usually results in many local
      minima that stop learning. The progress direction is guided and
      potential local minima are usually avoided. Even if they can't be
      avoided a back off is built in.

      Performance of Edit > Replace values on large grids is improved by a
      huge amount.

      Performance of Edit > Replace row and column names is also improved.

      Edit > Copy Example Rows on large grids is much faster.

      Tools > Extend Learning is about 100 times faster.

      Query > Query > Seek cycle is faster.

      The default target error has been changed to 0.01

      The minimum cycle count has been removed. Some networks only need one
      cycle.

      The "Stop after" and "Stop on cycle" controls override the Target
      error stop.

      Learning now ignores the Target error stop if any of the Validating or
      Score stops are set.

      The problems with the dialog title in Edit > Replace has been fixed.

      The problems with Edit > Replace row and column names have been fixed.

      The installer is at http://www.easynn.com/71049/ennplus.exe

      Registered users can use their current registration name and number.

      Other users need to email support@... for a temporary
      registration.

      This beta copy will run until the end of November 2005.

      --
      Steve Wolstenholme Neural Planner Software

      EasyNN-plus. The easy way to build neural networks.
      http://www.easynn.com
    • steve@easynn.com
      The first beta release of EasyNN-plus 7.0h is now available for testing. This is a full rebuild after moving from VC++ 6 to VC++ 8. Many source code changes
      Message 2 of 2 , Jan 13, 2006
      • 0 Attachment
        The first beta release of EasyNN-plus 7.0h is now available for
        testing.

        This is a full rebuild after moving from VC++ 6 to VC++ 8. Many source
        code changes have been necessary to replace the deprecated code with
        C++ standard, supported methods.

        Other changes are as follows:-

        1. If importing an XLS file, the first dialog now includes a check
        box to ask for example row types while they are being imported.

        2. Grids with many text cells now refresh much faster.

        3. Edit defaults have been changed so that the automatic move to the
        next cell can be either, if the value is edited or every time the edit
        dialog is closed.

        4. The Open defaults now include "Last edited cell" in the first
        mouse click "goto" radio buttons.

        5. The first mouse click "goto" is inhibited if a previous selection
        has used the keyboard.

        6. A History button has been added to the Information dialog. This
        shows the last twenty file open/save dates and times.


        The installer for beta 1 is at http://www.easynn.com/62303/ennplus.exe

        Registered users can use their current registration name and number.

        Other users need to email support@... for a temporary
        registration.

        This beta copy will run until the end of March 2006.

        --
        Steve Wolstenholme Neural Planner Software

        EasyNN-plus. The easy way to build neural networks.
        http://www.easynn.com
      Your message has been successfully submitted and would be delivered to recipients shortly.