No More Pesky Learning Rates
This addresses the need for automated hyperparameter tuning in machine learning, though it is incremental as it builds on existing adaptive optimization methods.
The paper tackles the problem of manually tuning learning rates in stochastic gradient descent by proposing a method that automatically adjusts multiple learning rates based on local gradient variations, achieving performance matching SGD with optimal settings on convex and non-convex tasks.
The performance of stochastic gradient descent (SGD) depends critically on how learning rates are tuned and decreased over time. We propose a method to automatically adjust multiple learning rates so as to minimize the expected error at any one time. The method relies on local gradient variations across samples. In our approach, learning rates can increase as well as decrease, making it suitable for non-stationary problems. Using a number of convex and non-convex learning tasks, we show that the resulting algorithm matches the performance of SGD or other adaptive approaches with their best settings obtained through systematic search, and effectively removes the need for learning rate tuning.