Options

Top  Previous 

 

 

The min/max values of the ranges of X and Y are taken by default from the .MMX file. You may want to alter these values on the screen in the Min/Max boxes to adjust the range of variations shown on the graph. However,  be careful not to make the range larger than the range with which the network was trained, especially if it was trained with clipping.

 

Other inputs of the network are set at a mean value for each corresponding input variable. One can alter these settings and enter new fixed value.

 

Restore Original Configuration option is available from the File menu. It allows to forget all changes made to the min/max/mean settings and to restore original values from the .MMX file.

 

The program automatically saves all current settings for a problem in a file called <problem>.CFG. The next time when you run the 3DGraphics programs it will remember all settings from the previous session.

 

If a contribution file <problem>.CXX is found in the data directory, then the network input variables are sorted in "Network inputs" list by their significance (contribution factors). This file is created by NeuroShell 2 for feedforward networks only in the Contribution module. When running a GRNN genetic adaptive net, individual contribution factors are saved directly in the .DEF file and they are used by the 3D Graphics program automatically.

 

The 3D picture has a feature which is called "Graphics Wizard". Point by mouse on the picture and click the right mouse button. A menu will appear providing full access to the picture properties. One can save the graph into BMP, WFM, or VCT formats. This allows importing of the generated picture into popular graphics design and publisher programs.