Table of Contents

SOM Tutorial

A. Creating the Network

  1. Use the following parameters to create your first SOM neural network using FuzzyCope/3:

    Network Architecture:
    Input Nodes 4
    Two Dimensions
    X 4
    Y 5

  2. Press the create button.

  3. Save the network as IrisSOM.wgt

  4. View the network


B. Training the network

  1. Open the training file that you formatted earlier - IrisSOM.trn

  2. View the training file

  3. Set the parameters for your Iris neural network:
    Learning Rate 0.1
    Neighbourhood 2 2
    Epochs 1000 then 5000

  4. Press the Train button to commence training

  5. Save the trained network as TIrisSOM.wgt


C. Network Evaluation

  1. From the Feature Map tab press the Map button to create the map dimensions.

  2. Open the recall file that you formatted earlier - IrisSOM.rcl

  3. From the Label/Recall tab press the Recall Button.

  4. Save the new output file (through the Recall menu).

  5. View the Output file.

  6. Press the Label button to label the network.

  7. Save the label file (through the Label menu).

  8. Compare your results with the desired outputs.

  9. Go back to the Feature Map tab and plot the map.


D. Additional Work

Try different combinations of:

  • Map Dimensions
  • Learning Rate
  • Neighbourhood parameters

    This page is maintained by Melanie Middlemiss mmiddlemiss@infoscience.otago.ac.nz
    Last modified on: 22/12/98.