2019/05/29 by Tomoya Murata, Taiji Suzuki, Murata, Tomoya +1
Computer Science · #Advanced Neural Network Applications #FOS: Computer and information sciences #FOS: Mathematics #Machine Learning (cs.LG) #Medical Image Segmentation Techniques #Optimization and Control (math.OC) #Stochastic Gradient Optimization Techniques
paper · pdf · doi:10.48550/arxiv.1905.12224
openalex publication_date 2019/05/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
A stochastic gradient method for synchronous distributed optimization is studied. For reducing communication cost, we particularly focus on utilization of compression of communicated gradients. Several work has shown that \itsparsified stochastic gradient descent method (SGD) with \iterror feedback asymptotically achieves the same rate as (non-sparsified) parallel SGD. However, from a viewpoint of non-asymptotic behavior, the compression error may cause slower convergence than non-sparsified SGD in early iterations. This is problematic in practical situations since early stopping is often adopted to maximize the generalization ability of learned models. For improving the previous results, we propose and theoretically analyse a sparsified stochastic gradient method with error feedback scheme combined with \itNesterov's acceleration. It is shown that the necessary per iteration communication cost for maintaining the same rate as vanilla SGD can be smaller than non-accelerated methods in convex and even in nonconvex optimization problems. This indicates that our proposed method makes a better use of compressed information than previous methods. Numerical experiments are provided and empirically validates our theoretical findings.