L4: Practical loss-based stepsize adaptation for deep learning [article]

Michal Rolinek, Georg Martius
2018 arXiv   pre-print
We propose a stepsize adaptation scheme for stochastic gradient descent. It operates directly with the loss function and rescales the gradient in order to make fixed predicted progress on the loss. We demonstrate its capabilities by conclusively improving the performance of Adam and Momentum optimizers. The enhanced optimizers with default hyperparameters consistently outperform their constant stepsize counterparts, even the best ones, without a measurable increase in computational cost. The
more » ... formance is validated on multiple architectures including dense nets, CNNs, ResNets, and the recurrent Differential Neural Computer on classical datasets MNIST, fashion MNIST, CIFAR10 and others.
arXiv:1802.05074v5 fatcat:6ikxfe5lg5c75bj7qwjtd7zrbe