used in binary classification problems
L(y, f(x)) = max(0, 1 - y * yhat)
- Where y is either +1 or -1
- Notice how if y is the same sign as yhat, the
1 - y * yhatterm becomes smaller- if this term is less than 0, the max function just makes the loss 0!
- So the more items that you classify correctly, the smaller the loss.