Weights
- The inputs can be modified by weights. (This also adds a similarity
to biological neural nets.)
- Now we can change the weights to change the system.
-
![Two Input Perceptron](perceptron.gif)
- Let's use a step function.
- The input output pairs are ((5,7), 1), ((3,10), 0), ((5,10),1), and
((5,2),0)
- If the threshold is 0, and the weights are both 1, the outputs are all
1.
- If we move the threshold to 12, the last ouput is 0, and the first
three are 1. We want the second to be 0.
- If we change the second weight to .5 and the threshold to 8.2, it gives
the correct result.