1 paper
Md Sakir Ahmed, Kumaresh Sarmah, Hemen Dutta
A widely held intuition in deep learning is that stochastic gradient descent (SGD) implicitly favors flat minima and that flat minima generalize better, but standard Euclidean meas…