Distributed Newton Methods for Deep Neural Networks

• Published in: Neural computation Vol. 30; no. 6; pp. 1673 - 1724
• Main Authors: Wang, Chien-Chih, Tan, Kent Loong, Chen, Chun-Ting, Lin, Yu-Hsiang, Keerthi, S. Sathiya, Mahajan, Dhruv, Sundararajan, S., Lin, Chih-Jen
• Format: Journal Article
• Language: English
• Published: One Rogers Street, Cambridge, MA 02142-1209, USA MIT Press 01.06.2018; MIT Press Journals, The
• Subjects: Artificial intelligence; Artificial neural networks; Communication; Convex analysis; Jacobi matrix method; Jacobian matrix; Letters; Machine learning; Mathematical analysis; Matrix algebra; Matrix methods; Methods; Neural networks; Neutrality; Newton methods; Nonlinear programming; Optimization; Synchronism; Training
• ISSN: 0899-7667; 1530-888X
• DOI: 10.1162/neco_a_01088

Deep learning involves a difficult nonconvex optimization problem with a large number of weights between any two adjacent layers of a deep structure. To handle large data sets or complicated networks, distributed training is needed, but the calculation of function, gradient, and Hessian is expensive. In particular, the communication and the synchronization cost may become a bottleneck. In this letter, we focus on situations where the model is distributedly stored and propose a novel distributed Newton method for training deep neural networks. By variable and feature-wise data partitions and some careful designs, we are able to explicitly use the Jacobian matrix for matrix-vector products in the Newton method. Some techniques are incorporated to reduce the running time as well as memory consumption. First, to reduce the communication cost, we propose a diagonalization method such that an approximate Newton direction can be obtained without communication between machines. Second, we consider subsampled Gauss-Newton matrices for reducing the running time as well as the communication cost. Third, to reduce the synchronization cost, we terminate the process of finding an approximate Newton direction even though some nodes have not finished their tasks. Details of some implementation issues in distributed environments are thoroughly investigated. Experiments demonstrate that the proposed method is effective for the distributed training of deep neural networks. Compared with stochastic gradient methods, it is more robust and may give better test accuracy.