Decentralized Gradient Tracking with Local Steps
This work addresses communication efficiency in decentralized machine learning training, offering a method to reduce overhead while handling data heterogeneity, though it appears incremental as it builds on existing gradient tracking techniques.
The paper tackles decentralized optimization with data heterogeneity by proposing K-GT, a novel gradient tracking algorithm that enables communication-efficient local updates, reducing communication overhead by a linear factor K and proving convergence on smooth non-convex functions.
Gradient tracking (GT) is an algorithm designed for solving decentralized optimization problems over a network (such as training a machine learning model). A key feature of GT is a tracking mechanism that allows to overcome data heterogeneity between nodes. We develop a novel decentralized tracking mechanism, $K$-GT, that enables communication-efficient local updates in GT while inheriting the data-independence property of GT. We prove a convergence rate for $K$-GT on smooth non-convex functions and prove that it reduces the communication overhead asymptotically by a linear factor $K$, where $K$ denotes the number of local steps. We illustrate the robustness and effectiveness of this heterogeneity correction on convex and non-convex benchmark problems and on a non-convex neural network training task with the MNIST dataset.