Quote:
Originally Posted by sninoz
If w_0 can be chosen to achieve the desired boundary, is it not the case that there is ambiguity in the answers? In particular, can't switching the signs of w_1 and w_2 work (as long we also have the appropriate change in sign for w_0, which is allowed since w_0 can be 'chosen to achieve the desired boundary')?

The sign of
and
cannot be compensated by changing the value of
. All these weights enter the formula that determines the function being implemented in a specific way.