Previous Next

*   Using the Neural Networks Window
A Neural Networks window displays a topological map of neurons, represented as rectangles. Every neuron has a minimum of two and a maximum of six neighboring neurons. Spectra are displayed as symbols or numbers within neurons.
The status bar in a Neural Network window displays the lattice dimension and how many spectra have been classified (the total number of spectra from all groups).
v
To show distances between neurons
Click the Show Distances Between Neurons button, .
To show the approximate distances between neighboring neurons, the application applies different thicknesses to their borders.
You can use information about neuron distances when classifying an unknown spectrum. The distances show how a neuron that has been activated by an unknown correlates with neighboring neurons. The distances can also suggest that spectra from a common neuron might have been activated by multiple, close neurons.
Although neurons are displayed in a regular 2‑D lattice, the actual Euclidian distances between neurons vary. The line thickness is proportional to the distance between immediate neighbors. The thinner the line, the closer together are the neurons.
Note  Spectra that activate the same neuron belong to the same classification pattern. This means that all spectra drawn inside a neuron box are equal for classification purposes and their graphical positions within a neuron are irrelevant.
v
To scale the lattice
1.
Right-click the Neural Networks window and choose Zoom > zoom option from the shortcut menu.
Click Zoom to Box to describe a rectangle that enlarges to fill the window.
Click Zoom In to display less of the neural networks lattice at a larger scale.
Click Zoom Out to display more of the neural networks lattice at a smaller scale.
2.
v
To use the zoom and pan features
Click the Hand button, , in the Mass Frontier toolbar and drag the hand to pan horizontally or vertically.
Similarly, hold down the H keyboard key or the Spacebar to change the cursor to the hand.
Click the Zoom Out button, , in the Mass Frontier toolbar.
Click the Zoom In button, , in the Mass Frontier toolbar.
Click the Zoom to Box button, , in the Mass Frontier toolbar and describe a rectangle.
Similarly, hold down the Z keyboard key and describe a rectangle.
Click the Zoom Reset button, , in the Mass Frontier toolbar to return to full scale.
Click the Zoom Undo button, , in the Mass Frontier toolbar to undo the last zoom operation.
Click the Zoom Redo button, , in the Mass Frontier toolbar to restore the last zoom operation.
v
To open a neural networks file
1.
Click the Open button, , in the Mass Frontier main toolbar and choose Neural Networks.
–or–
Choose File > Open > Neural Networks from the Mass Frontier main menu.
The Open Neural Networks dialog box opens.
2.
The lattice for the selected file opens in a new Neural Networks window.
v
To save neural networks to a file
1.
Click the Save button, , in the Mass Frontier main toolbar.
–or–
Choose File > Save from the Mass Frontier main menu.
The Save Neural Networks dialog box opens.
2.
The Mass Frontier application saves neural networks as graphics, without the possibility of recalling the original spectra, displayed as symbols. To access the spectra from a lattice, save the spectra to records in the Database Manager window as described in the Spectra Classifier chapter. In this case, regenerate the classification from the original data set that was saved.
For additional information about saving spectra to records, see Using Records in the Database Manager Window or Organizing Spectra.


Previous Next
Related Topics:
  Generating a Neural Networks Module
  Accessing Spectra from a Neural Networks Window
Copyright 1998 - 2013 HighChem Ltd., Slovakia