Gradient Descent Finds Global Minima of Deep Neural Networks — mentioned by mlthoughts2018

Sat, 08 Dec 2018 22:03:07 GMT

I suspect actually there is some connection between optimal batch size and the proofs of this paper regarding how deep NN will provably converge to zero training loss with gradient descent (not stochastic gradient descent). https://arxiv.org/abs/1811.03804

Paper and discussion history. A mention is not an endorsement.

Gradient Descent Finds Global Minima of Deep Neural Networks — HN submission

Mon, 12 Nov 2018 15:17:58 GMT

Gradient Descent Finds Global Minima of Deep Neural Networks

133 points; 57 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Gradient Descent Finds Global Minima of Deep Neural Networks · HackerPapers discussions

Gradient Descent Finds Global Minima of Deep Neural Networks — mentioned by mlthoughts2018

Gradient Descent Finds Global Minima of Deep Neural Networks — HN submission