next up previous
Next: Validity Testing Up: Results Previous: Results

simulation

  
Figure 12: An Example of window display for ANN training and rule eliciting

 

  
Figure 13: An example of intermediary rules displayed by the rule extraction algorithm; these rules set connections between the input layer and the hidden layer, and separately between the hidden layer and the output; from this stage, rules have to be integrated so that the direct input-output connection can be established.

 

  
Figure 14: An example of redundant rules displayed by the rule extraction algorithm, after the integration; these rules are extracted from the intermediary rules, and lead to redundancy (multiple appearance of the same rule); in order to eliminate this redundancy, a further step of deleting extra rules has to be done.



Alexandra Cristea
Tue Feb 9 20:20:27 JST 1999