Dissecting Adam: The Sign, Magnitude and Variance of Stochastic Gradients
arXiv:1705.07774
Abstract
The ADAM optimizer is exceedingly popular in the deep learning community. Often it works very well, sometimes it doesn't. Why? We interpret ADAM as a combination of two aspects: for each weight, the update direction is determined by the sign of stochastic gradients, whereas the update magnitude is determined by an estimate of their relative variance. We disentangle these two aspects and analyze them in isolation, gaining insight into the mechanisms underlying ADAM. This analysis also extends recent results on adverse effects of ADAM on generalization, isolating the sign aspect as the problematic one. Transferring the variance adaptation to SGD gives rise to a novel method, completing the practitioner's toolbox for problems where ADAM fails.
Presented at the 35th International Conference on Machine Learning (ICML), 2018
References in corpus (6)
- Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
- ADADELTA: An Adaptive Learning Rate Method
- SAGA: A Fast Incremental Gradient Method With Support for Non-Strongly Convex Composite Objectives
- New insights and perspectives on the natural gradient method
- No More Pesky Learning Rates
- Probabilistic Line Searches for Stochastic Optimization