Learning from Explanations