Well,this is surprising. Paper doesn't mention image recognotion and voice recognition at all. Is this because they are not interesting for new research or beca

Well,this is surprising. Paper doesn't mention image recognotion and voice recognition at all. Is this because they are not interesting for new research or because in theese applications results were worse than that of gradient methods?