attrasoft

Attrasoft

Attrasoft PolyNet 
for Windows 95/98
Version 4.0 (7/1999)
About

New

Image Search Engine

Products

Services

Order

Shareware/Demos

Future Products

Links

Search This Site


All Attrasoft products are Y2K Compliant!

6. Command Reference and Toolbar

6.1 File
6.2 Edit
6.3 Data
6.4 Decimal
6.5 Binary
6.6 Examples
6.7 Window
6.8 Help
6.9 Toolbar and Status Bar

6.   Command Reference and Toolbar

6.1   File

File/New

    Use the "New" command to create an empty file window. This window will be used for editing the data file.
File/Open
Use the "Open" command to open an existing file. 
     
File/Close
    Use the "Close" command to close an existing file. 
File/Save
    Use the "Save" command to immediately store the contents of the current window to a file while leaving the window active. "Save" can also be used to save your most recent change to a disk.
File/Save As
    Use the "Save As" command to immediately store the contents of the current window to a file that you specify. If a window is untitled, "Save As" should be used to save your most recent changes to the disk.
File/Print Preview
    Use the "Print Preview" command to preview before printing.
File/Print 
    Use the "Print" command to print the contents of a data file window.
File/Print Setup
    Use the "Print setup" command to set up the printer before you print.
File/Exit
    Use the "Exit" command to leave PolyNet for Windows 95/98. 
6.2   Edit

Edit/Undo

    Use the "Undo" command to reverse the most recent edit-actions. This command can only apply to editing, not training.
Edit/Cut
    Use the "Cut" command to cut the contents of the selected text. The Cut command also copies the contents of the selected text to the Windows 95/98 Clipboard. 
Edit/Copy
    Use the "Copy" command to copy the contents of the selected text to the Windows 95/98 Clipboard. 
Edit/Paste
    Use the "Paste" command to place the contents of the Windows 95/98 Clipboard into a selected position.
Edit/Clear All
    Use the "Clear All" command to delete all contents of a data file.
Edit/Delete
    Use the "Delete" command to delete the contents of the selected text.
Edit/Find
    Use the "Find" command to find a pattern in the data files.
Edit/Replace
Use the "Replace" command to replace one pattern in the data file by another.
Edit/Next
    Use the "Next" command to repeat the last Find or Replace.
Edit/Font
    Use the "Font" command to replace the current font.
6.3   Data

Data/Link

Use the "Data/Link" command to assign the files used by PolyNet. To assign file names, click the command and a dialog box will be presented. By default, the PolyNet will select the following names:

Training file: example1a.txt
Recognition file: example1b.txt
Output file: example1c.txt

Data/Line Breaker Use the "Data/Line Breaker " command to set line breakers for the output data file. If the output format is not set, the output will be printed one vector per line. This can be changed by inserting line breakers. 

If you want to insert line breakers, you must specify the position of the line breakers. PolyNet allows you to insert 4 line breakers.

Suppose you want the output to look like this:

111
0000
11111
000000
1111111

then, click: 'Data/Line Breaker' and specify:

Line 2 starts at 3
Line 3 starts at 7
Line 4 starts at 12
Line 5 starts at 18.

PolyNet allows the output neurons to be printed in five lines. However, if line 3 and line 4 have an equal length, all remaining lines also have the same length. For example, to change the output to:

1111
000
111
000
111
000
111
000
1,

Click: 'Data/Line Breaker' and specify:

Line 2 starts at 4
Line 3 starts at 7
Line 4 starts at 10
Line 5 starts at 13

Data/Symmetry Use the "Data/Symmetry " command to set symmetry for the training and retraining files. The symmetries only apply to the pattern-neurons. In general, the class-neurons do not have any symmetry. 

To specify the symmetry, you have to specify:

    • Where is the symmetry ?
    • What symmetry ?
Where

The position of a pattern and the dimension tell "where":

Position: where does it start, and where does it end:
Dimension: 1-dimension or 2-dimension. 


Select: "Data/Symmetry" to specify:

where symmetry starts
where symmetry ends
the x-dimension of the symmetry
the y-dimension of the symmetry


If you do not specify the symmetry, the default values are -2, meaning no symmetry. Please refer to the last section for the position of a pattern. 

What

To specify a symmetry, check the necessary boxes:
 

x-translation symmetry
y-translation symmetry
x-scaling symmetry
y-scaling symmetry
z- rotation symmetry
Data/Input Use the "Data/Input" command to open the training and recognition files directly without specifying the file name. This command will open the input file specified by "Data/Link".  Data/Output Use the "Data/Output" command to open the output file directly without specifying the file name. This command will open the output file specified by "Data/Link".  Data/Test Use the "Data/Test" command to test the training and recognition files. If the error is not corrected, the PolyNet will not run. 6.4   Decimal

Decimal/Distribution

    Use the "Decimal/Distribution" command to produce a class distribution. The recognition patterns are printed first in the output file, followed by all possible outputs. Each possibility is also associated with a relative probability. The one with the largest probability is the preferred answer. Finally, the preferred answer is printed again.
Decimal/Classification
    Use the "Decimal/Classification" command to get a classification of the recognition patterns. 
Decimal/1N1C Distribution
    This command is similar to but faster than "Decimal/Distribution" command. 1-neuron-for-1-class representation must be used, i.e. among class neurons, one of the neurons is '1' - '9'; all others are '0':

    100 ... 00
    010 ... 00
    001 ... 00
    ...
    000 ... 10
    000 ... 01

    Here each "1" represents a class classification.

Decimal/1N1C Classification
    This command is similar to but faster than "Decimal/Classification" command. 1-neuron-for-1-class representation must be used, i.e. among class neurons, one of the neurons is '1' - '9'; all others are '0':

    100 ... 00
    010 ... 00
    001 ... 00
    ...
    000 ... 10
    000 ... 01

    Here each "1" represents a class classification.

Decimal/Train
    Use the "Decimal/Train" command to train the network. Training will first delete all of the old connections, thus starting with no synaptic connections. The results of training will not be saved for future use because training does not take very long. 
Decimal/Retrain
    Use the "Decimal/Retrain" command to retrain the network. Retraining can not take place until training has been done first. You can have as many retraining as you wish. Every retraining looks for data from the training file. To offer fresh data to the PolyNet for each retraining, you have to change the training file. 
Decimal/Recognitition|Distribution
    Use the "Decimal/Recognition|Distribution" command to produce a class distribution, if the neural network is trained first. 
Decimal/Recognition|Classification
    Use the "Decimal/Recognition|Classification" command to get a classification of the recognition patterns, if the neural network is trained first. 
6.5   Binary

Binary/Distribution

    Use the "Binary/Distribution" command to produce a class distribution. The recognition patterns are printed first in the output file, followed by all possible outputs. Each possibility is also associated with a relative probability. The one with the largest probability is the preferred answer. Finally, the preferred answer is printed again.
Binary/Classification
    Use the "Binary/Classification" command to get a classification of the recognition patterns. 
Binary/1N1C Distribution
    This command is similar to but faster than "Binary/Distribution" command. 1-neuron-for-1-class representation must be used, i.e. among class neurons, one of the neurons is '1'; all others are '0':

    100 ... 00
    010 ... 00
    001 ... 00
    ...
    000 ... 10
    000 ... 01

    Here each "1" represents a class classification.

Binary/1N1C Classification
    This command is similar to but faster than "Binary/Classification" command. 1-neuron-for-1-class representation must be used, i.e. among class neurons, one of the neurons is '1'; all others are '0':

    100 ... 00
    010 ... 00
    001 ... 00
    ...
    000 ... 10
    000 ... 01

    Here each "1" represents a class classification.

Binary/Local Minimum(1)
    Use the "Binary/Local Minimum(1)" command to complete a pattern. This option simulates the network for a number of steps, but stops way short of the main command, Binary/Distribution. The most likely solution is printed in the output data file. 
Binary/Local Minimum(5)
    Similar to command "Binary/Local Minimum(1)" except this command starts from 5 random configurations and lists 5 local minimum.
Binary/Train
    Use the "Binary/Train" command to train the network. Training will first delete all of the old connections, thus starting with no synaptic connections. The results of training will not be saved for future use because training does not take very long. 
Binary/Retrain
    Use the "Binary/Retrain" command to retrain the network. Retraining can not take place until training has been done first. You can have as many retraining as you wish. Every retraining looks for data from the training file. To offer fresh data to the PolyNet for each retraining, you have to change the training file. 
Binary/Recognition|Distribution
    Use the "Binary/Recognition|Distribution" command to produce a class distribution, if the neural network is trained first. 
Binary/Recognition|Classification
    Use the "Binary/Recognition|Classification" command to get a classification of the recognition patterns, if the neural network is trained first. 
     
6.6   Examples

Example/Shifter

Use the "Example/Shifter" command to open a pop up menu, the shifter menu. The shifter menu allows you to study the shifter problem. Example/Shifter/Data Use the "Example/Shifter/Data" command to generate training and recognition data for the Shifter problem. By default, this command generates 500 training patterns and 50 recognition patterns for the 15-digit shifters. This command also sets up the neural net for you. Example/Shifter/Complete Pattern Use the "Example/Shifter/Pattern" command to generate the recognition data for the 15-Digit Shifter problem. A classification and a part of a pattern are given, the neural net is asked to complete the pattern. To complete a pattern, please first click "Example/Shifter/Data" to generate the training file, then click this command to generate the recognition file.  Example/Shifter/Parameter  Use the "Example/Shifter/Parameter" command to see and change the parameter of the Shifter problem. The parameters are :
    • The number of digits for the N-digit Shifter;
    • The number of training patterns; and
    • The number of recognition patterns. 
The default values for the shifter problem are: N = 15 (33 neurons), 500 training-patterns, and 50 recognition-patterns. Example/Double Shifter
Example/Double Shifter/Data
Example/Double Shifter/Complete Patterns
Example/Double Shifter/Parameter Similar to the commands of the Shifter except they are for the Double Shifter problems. Example/Triple Shifter
Example/Triple Shifter/Data
Example/Triple Shifter/Complete Patterns
Example/Triple Shifter/Parameter Similar to the commands of the Shifter except they are for the Triple Shifter problems. Example/Quadruple Shifter
Example/Quadruple Shifter/Data
Example/Quadruple Shifter/Complete Patterns
Example/Quadruple Shifter/Parameter Similar to the commands of the Shifter except they are for the Quadruple Shifter problems. Example/1D Segment Counter Use the "Example/1D Segment Counter " command to open a pop up menu, the 1D-Segment-Counter menu. This example counts the number of segments. For example, a string:

111100111 

has two segments: 1111 and 111. The class-vector has 40 digits:

10000000... : no segment 
01000000... : 1 segment 
00100000... : 2 segments 
00010000... : . . .

This problem counts from 0 to 38 segments; if the data has more than 38 segments, it is class 00 . . . 001. The default pattern-vector has 120 digits, therefore the input-vector has 160 neurons. The data is generated as follows: the probability of '1' - '9' is 34% and that of '0' is 66%. The segment length is 4 or higher. The default values are: 

120 pattern-neurons and 40 class-neurons (160 neurons), 
500 training-patterns, and 
50 recognition-patterns.

Example/1D Segment Counter/Data Use the "Example/1D Segment counter/Data" command to generate training and recognition data. This command generates 500 training patterns and fifty recognition patterns.  Example/1D Segment Counter/Parameter  Use the "Example/1D Segment Counter/Parameter" command to see and change the parameter. The parameters are :
    • The number of digits for 1D Segment Counter;
    • The number of patterns; and
    • The number of patterns. 
The default values for the shifter problem are: N = 120 (160 neurons), 500 training patterns, and 50 recognition patterns. Example/2D Scaling Symmetry The "2D scaling Symmetry" has 40x40 images of 16 different airplanes as training data. In the recognition data, there are 10 images of the airplane at different scales. Users can add their own data at the end of the data files. Make sure you test the data file format before running. To check the format, click "Data/Test".  Example/2D Rotation + Translation The "Example/2D Rotation + Translation" command opens two files: training and recognition files. The training-data-file has 40x40 images of 16 different airplanes. In the recognition data, there are 11 images of the airplanes at various angles. Users can add their own data at the end of the recognition/training files. Make sure you test the data file format before running. To check the format, click "Data/Test".  Example/2D Rotation + Scaling The "Example/2D Rotation + Scaling" command opens two files: training and recognition files. The training-data-file has 40x40 images of 3 different airplanes. In the recognition data, there are 10 images of the airplanes at various angles and various sizes. Users can add their own data at the end of the recognition/training files. Make sure you test the data file format before running. To check the format, click "Data/Test".  Example/5x7 Character  Use the "Example/5x7 Character" command to create the '5 by 7' character recognition problem. The command generates both the training and recognition data for the '5 by 7' character recognition problem. Fifty or so training-patterns and 10 recognition patterns will be generated. The command also initializes the PolyNet for the character recognition problem.  Example/8x8 Character  Use the "Example/8x8 Character" command to create the 8 by 8 character recognition problem. Example/19x19 Character (1,023 classes)  The training file contains all 1,023 classes. The neural net is organized accordingly: the class-vector has 1,026 neurons and the pattern-vector has 19 * 19 = 361 neurons, giving a total of 1,026 + 361 = 1,387 neurons. The reason to choose 1,026 neurons instead of 1,023 is that the output-vector will be printed 19-digits per line and 1,387 neurons will fit into 73 lines. 

The recognition file contains several recognition-patterns generated randomly. If you want more data, just click this command again; each time, different recognition-patterns will be generated.

Example/19x19 Character (4,095 classes)  Similar to the "Example/19x19 Character (4,095 classes)" except 4,095 characters are generated for the training-data-file. Example/32x32 Character  There are 127 classes in the training file. The neural net is organized accordingly: the class-vector has 128 neurons and the pattern-vector has 32 * 32 = 1,024 neurons, giving a total of 1,024 + 128 = 1,152 neurons. The reason to choose 128 neurons instead of 127 is that the output-vector will be printed 32-digits per line and 1,152 neurons will fit into 36 lines.  Example/50x50 Binary Image Recognition  The data files are img50x50.trn and img50x50.rec. Images of a face, a rabbit, a boat, . . . are used to train the neural net. Users can add their own data to the end of the recognition and training files. Make sure you check the data file format before running. To check the format, click "Data/Test".  Example/100x100 Binary Image Recognition  Similar to the "Example/50x50 Binary Image Recognition" command. The data files are img100x100.trn and img100x100.rec. Images used to train the neural net are:

10000000000 F104
01000000000 B-57
00100000000 Mirage
00010000000 F-105
00001000000 MIG

Users can add their own data at the end of the recognition and training files. Make sure you test the data file format before running. To check the format, click "Data/Test". 

Example/Math Function /245 Class  Use the "Example/Math Function /245 Class" command to generate an example defined by:

y1 = ( x1 + x2 + x3 ) Mod 7; 
y2 = x1; 
y3 = x2 Mod 5; 

This problem has 245 classes. The data is generated as follows: assume initially that (x1, x2, x3) is in state (1 0 0), then it will generate (1 1 0). Now we will use (x1, x2, x3) = (1 1 0) and it will generate (2 1 1), ...

There are 69 rows of data, which are divided into two parts: the first 64 rows will be the training file and the last 5 rows will be the recognition file.

Example/Math Mapping /3675 class  Use the "Example/Math Mapping /3675 class" command to generate an example defined by:

y1 = ( x1 + x2 + x3 ) Mod 7; 
y2 = x1; 

y3 = x2 Mod 5; 60%
y3 = ( x2 + 1 ) Mod 5; 40% 

y4 = ( x4 + 1 ) Mod 5; 60% 
y4 = ( x4 + 2 ) Mod 5; 40% 

y5 = x4 Mod 3.

We will generate the data as follows: assume initially that (x1, x2, x3, x4, x5) is in state (1 0 0 0 0), then it will generate (1 1 0 1 0). Now we will use (x1, x2, x3, x4, x5) = (1 1 0 1 0) and it will generate an answer (2 1 2 3 1), ...

There are 9999 rows of data, which is divided into two parts: first 9994 rows will be the training file and the last 5 rows will be the recognition file.

Example/Cancer Database Use the "Example/Cancer Database" to open the Cancer Database example in chapter 2.  Example/Nursery School Database Use the "Example/Nursery School Database" to open Nursery School Database example in chapter 5. 6.7   Window
     
Window/Tile
    Use the "Window/Tile" command to arrange your open data Windows so that all Windows are visible and roughly equally sized across the screen.
Window/Cascade
    Use the "Window/Cascade" command to arrange your open data Windows so that all Windows are roughly equally sized and layered, leaving only their title bar and left border visible.
Window/Arrange Icons
    Use the "Window/Arrange Icons" command to arrange the icons of data Windows. The icons for data Windows are created when you minimize the data file Windows.
Window/Close All
    Use the "Window/Close All" command to close all of your opened data Windows. 
     
6.8   Help

Help/Contents

To use on-line help, click Help|Contents or the Help-Button. The on-line help has three parts:
    • Commands;
    • Toolbar; and 
    • Procedures. 
The commands are given in this chapter. The toolbar has 19 buttons. The list of the buttons is given in the next section. Under the Procedure, you will see:
    • Introduction
    • Total Operation in 2-4 Clicks
    • User's Guide
    • Symmetry
    • Examples
Help/Using Help
The Help|Using Help displays the help index.
Help/Note Use the "Help/Note" command to open your private notes about the software. You can write your own notes here.  Help/About Use the "Help/About" command to access the Attrasoft "About" information box. The "About" dialogue box contains information like the version number, the release date, the company address and the copyright information. 6.9   Toolbar and Status Bar PolyNet toolbar has the following buttons. They are listed in the following order: File/New;
File/Open;
File/Save;
Data/Link
Data/Input
Data/Output
Decimal/Classification
Decimal/Distribution
Decimal/Train
Decimal/Retrain
Decimal/Recognition|Classification
Decimal/Recognition|Distribution
Edit/Cut;
Edit/Copy;
Edit/Paste;
Edit/Undo;
File/Print;
File/Print Preview.
Help/Contents


You can identify these buttons by putting (not clicking) the mouse on top of a button and reading the status bar.

The status bar explains all the commands in the menu bar and all the buttons on the toolbar.

 
Online Order or if you prefer, Online Fax order: 
PolyNet for Windows 95/98 
Version 4.0 ($124.99 + $5 US Shipping and Handling)

Mail Order: $129.99 (S&H included)

PolyNet 4.0
Attrasoft, Attn.: Gina
P. O. Box 13051
Savannah, GA. 31406, USA
Products New Lottery Stock jpg/gif Order Order Help Demos
Please join Attrasoft Discussion Group.
Send questions or comments to:
webmaster@attrasoft.com
or contact us at: 
          Attrasoft, P. O. Box 13051, Savannah, GA. 31406, USA
Copyright © 1998 Attrasoft, Inc. All rights reserved.