SLowcal-SGD: Slow Query Points Improve Local-SGD for Stochastic Convex Optimization
This addresses the challenge of efficient distributed optimization for heterogeneous data, offering a novel improvement over existing baselines.
The paper tackles the problem of distributed learning with heterogeneous data distributions by introducing SLowcal-SGD, a local update method that provably outperforms Minibatch-SGD and Local-SGD through a slow querying technique to reduce bias from local updates.
We consider distributed learning scenarios where M machines interact with a parameter server along several communication rounds in order to minimize a joint objective function. Focusing on the heterogeneous case, where different machines may draw samples from different data-distributions, we design the first local update method that provably benefits over the two most prominent distributed baselines: namely Minibatch-SGD and Local-SGD. Key to our approach is a slow querying technique that we customize to the distributed setting, which in turn enables a better mitigation of the bias caused by local updates.