Due Tue July 12, 2005
Here is the problem we talked about today. We had a network that:
For p1 = [1 1], p2 = [1 -1], and p3 = [-1 1] produces 1 as output and
For p4 = [-1 -1] produces 0 as output
We want to try two different approaches:
1) We take w = [-1 1] as the weight and will try to find a bias that works for all of the inputs correctly
2) We take w = [1 -1] as the weight and will try to find a bias that works for all of the inputs correctly.
We have gone through in class for w = [1 1]. Now you will solve it for two different new ws. Is it possible to use the new ws and solve the problem?