2 ms·
Of course there is some work in that direction- but it is not the majority of the field, not by a long shot. The advances you list are interesting and they are
by YeGoblynQueenne 9y ago
Of course there is some work in that direction- but it is not the majority of the field, not by a long shot. The advances you list are interesting and they are possibly the building blocks of a larger theory, but they remain few and far between- occasional, disconnected contributions on the side of the main work in deep learning. Without a concentrated effort to build on them, rather than chase records, we will not see a "calculus of deep learning" any time soon.