- how neural networks can compute non-linear hypothesis

- example of negation
- put large number to negate each other

- 0 0 = - 30(1)
- 0 1 = 10(0)
- 1 0 = 10 (0)
- 1 1 = 10 (0)

- xnor need non-linear decision boundaries
- xnor outputs 1 when either x1 and x2 are the same

- this is why neural networks can compute more and more complicated function