Optimal mini-batch and step sizes for SAGA
This work provides incremental improvements for researchers and practitioners using stochastic variance-reduced methods like SAGA, by offering practical guidelines for parameter tuning to enhance convergence efficiency.
The paper tackles the problem of determining optimal step sizes and mini-batch sizes for the SAGA algorithm, a variance-reduced gradient method, by deriving closed-form expressions for the expected smoothness constant and showing that total complexity decreases linearly with mini-batch size up to an optimal value, with results competitive with numerical grid searches.
Recently it has been shown that the step sizes of a family of variance reduced gradient methods called the JacSketch methods depend on the expected smoothness constant. In particular, if this expected smoothness constant could be calculated a priori, then one could safely set much larger step sizes which would result in a much faster convergence rate. We fill in this gap, and provide simple closed form expressions for the expected smoothness constant and careful numerical experiments verifying these bounds. Using these bounds, and since the SAGA algorithm is part of this JacSketch family, we suggest a new standard practice for setting the step sizes and mini-batch size for SAGA that are competitive with a numerical grid search. Furthermore, we can now show that the total complexity of the SAGA algorithm decreases linearly in the mini-batch size up to a pre-defined value: the optimal mini-batch size. This is a rare result in the stochastic variance reduced literature, only previously shown for the Katyusha algorithm. Finally we conjecture that this is the case for many other stochastic variance reduced methods and that our bounds and analysis of the expected smoothness constant is key to extending these results.