Backpropagation .*;
4x6x14 Network Example
This example uses 4 input, 6 hidden and 14 output nodes to classify 14 patterns.
In the Java version, I've introduced a noise factor which varies the original input a little, just to see how much the network can tolerate.
Example Results 1
It generally works pretty well. The noise factor setting here is 0.45. The correct output should go in order. The noise causes the net to predict input #1 as #8.
Network is 100.0% correct. Test network against original input: 1.0 1.0 1.0 0.0 Output: 0 1.0 1.0 0.0 0.0 Output: 1 0.0 1.0 1.0 0.0 Output: 2 1.0 0.0 1.0 0.0 Output: 3 1.0 0.0 0.0 0.0 Output: 4 0.0 1.0 0.0 0.0 Output: 5 0.0 0.0 1.0 0.0 Output: 6 1.0 1.0 1.0 1.0 Output: 7 1.0 1.0 0.0 1.0 Output: 8 0.0 1.0 1.0 1.0 Output: 9 1.0 0.0 1.0 1.0 Output: 10 1.0 0.0 0.0 1.0 Output: 11 0.0 1.0 0.0 1.0 Output: 12 0.0 0.0 1.0 1.0 Output: 13 Test network against noisy input: 1.4 1.4 1.0 0.1 Output: 0 1.1 1.2 0.4 0.4 Output: 8 0.1 1.2 1.2 0.3 Output: 2 1.3 0.3 1.3 0.4 Output: 3 1.1 0.1 0.2 0.2 Output: 4 0.2 1.2 0.3 0.3 Output: 5 0.3 0.0 1.3 0.4 Output: 6 1.1 1.3 1.2 1.4 Output: 7 1.3 1.0 0.3 1.4 Output: 8 0.0 1.2 1.4 1.1 Output: 9 1.2 0.2 1.4 1.2 Output: 10 1.3 0.1 0.3 1.4 Output: 11 0.2 1.2 0.3 1.4 Output: 12 0.3 0.4 1.4 1.4 Output: 13
Example Results 2
It can take a little noise, but not too much. The noise factor here is 0.58.
Network is 100.0% correct. Test network against original input: 1.0 1.0 1.0 0.0 Output: 0 1.0 1.0 0.0 0.0 Output: 1 0.0 1.0 1.0 0.0 Output: 2 1.0 0.0 1.0 0.0 Output: 3 1.0 0.0 0.0 0.0 Output: 4 0.0 1.0 0.0 0.0 Output: 5 0.0 0.0 1.0 0.0 Output: 6 1.0 1.0 1.0 1.0 Output: 7 1.0 1.0 0.0 1.0 Output: 8 0.0 1.0 1.0 1.0 Output: 9 1.0 0.0 1.0 1.0 Output: 10 1.0 0.0 0.0 1.0 Output: 11 0.0 1.0 0.0 1.0 Output: 12 0.0 0.0 1.0 1.0 Output: 13 Test network against noisy input: 1.2 1.2 1.3 0.4 Output: 0 1.5 1.0 0.1 0.5 Output: 8 0.4 1.1 1.2 0.6 Output: 9 1.0 0.5 1.1 0.2 Output: 3 1.5 0.1 0.1 0.2 Output: 4 0.5 1.4 0.1 0.5 Output: 5 0.5 0.4 1.5 0.0 Output: 3 1.5 1.5 1.2 1.0 Output: 7 1.3 1.5 0.3 1.4 Output: 8 0.4 1.3 1.4 1.5 Output: 9 1.3 0.3 1.3 1.4 Output: 10 1.2 0.3 0.3 1.3 Output: 11 0.1 1.2 0.4 1.2 Output: 9 0.1 0.1 1.1 1.1 Output: 13