Week3 y=1 decision boundary

[flyunicorn]

As compared to the second screenshot, why in first screenshot, y=1 is the area inside instead of outside?

[dtonhofer]

It’s arbitrary, you are free to choose however you want. It is all about adding a negative sign or not.

One way:

1. Every x for which f(x) > 0 is declared “belonging to class 0”
2. Every x for which f(x) < 0 is declared “belonging to class 1”

Another way

1. Every x for which f(x) < 0 is declared “belonging to class 0”
2. Every x for which f(x) > 0 is declared “belonging to class 1”

Decision boundaries may not even have an “inside” or “outside”, for example, they may cut the the plane into two half-planes.

[TMosh]

It depends on the sign of the weights learned from the training set