Demonstration 2a Loading a File and Training with a CSV File - adam-nnl/Erudite GitHub Wiki

For this demo we will train a simple backpropogation neural network with the XOR logic truth table.

  1. Go into your Erudite-v1.1.0-Full_Suite folder and launch ‘Erudite-v1.1.0.jar

  2. After the program loads click the ‘Browse…‘ button in the ‘File Settings‘ section

  3. Navigate into the ‘ANNeML Examples‘ directory and select/open ‘logicTest-alllogsig_ANNeML.xml

  4. Go to the ‘Training Settings‘ section and check the box for ‘Enable Training

  5. Click the now-enabled button ‘Import Training Set

  6. Navigate to the ‘Training Sets‘ directory and select/open ‘XOR_TS-logsig.csv

  7. Select the option for ‘Avg. error< 0.10

  8. Click the ‘Train NNet‘ button. The neural network with now go through as many training cycles as it needs to to “learn” the training set.

  9. Once training has completed uncheck the ‘Enable Training‘ box to test your results with the next demonstration

Continue to Demonstration 2b