Accumulated Gradient Normalization
This addresses optimization challenges in distributed machine learning, though it appears incremental as it builds on existing asynchronous methods.
The paper tackles the instability in asynchronous data parallel optimization by introducing a distributed optimizer that normalizes accumulated gradients, mitigating parameter staleness and improving convergence rates compared to methods like asynchronous EASGD and DynSGD.
This work addresses the instability in asynchronous data parallel optimization. It does so by introducing a novel distributed optimizer which is able to efficiently optimize a centralized model under communication constraints. The optimizer achieves this by pushing a normalized sequence of first-order gradients to a parameter server. This implies that the magnitude of a worker delta is smaller compared to an accumulated gradient, and provides a better direction towards a minimum compared to first-order gradients, which in turn also forces possible implicit momentum fluctuations to be more aligned since we make the assumption that all workers contribute towards a single minima. As a result, our approach mitigates the parameter staleness problem more effectively since staleness in asynchrony induces (implicit) momentum, and achieves a better convergence rate compared to other optimizers such as asynchronous EASGD and DynSGD, which we show empirically.