On the expected running time of nonconvex optimization with early stopping

no code implementations • 25 Sep 2019 • Thomas Flynn, Kwang Min Yu, Abid Malik, Shinjae Yoo, Nicholas D'Imperio

This work examines the convergence of stochastic gradient algorithms that use early stopping based on a validation function, wherein optimization ends when the magnitude of a validation function gradient drops below a threshold.

Layered SGD: A Decentralized and Synchronous SGD Algorithm for Scalable Deep Neural Network Training

no code implementations • 13 Jun 2019 • Kwangmin Yu, Thomas Flynn, Shinjae Yoo, Nicholas D'Imperio

The efficiency of the algorithm is tested by training a deep network on the ImageNet classification task.