News for hackers. Stuff that matters.
Re:Why deep learning works even though it shouldn’t by muppet_frog on Tuesday October 20, 2020 @05:29PM (#24842534)
This paper makes the points that it's the saddles and not local minima that are the problem: https://arxiv.org/abs/1406.2572 It was the basis for adding 'momentum' to optimizers - so that you could skate across the saddles.

There are no comments on this comment yet.