A quick idea.
I assume many functions that machine learning tries to mimic has divisions in them. With division comes to possibility for division by zero. However I think there is a relation between high risk and high reward in machine learning also.
One such reward could be faster convergence.