A Convergence Analysis of Gradient Descent for Deep Linear Neural Networks

10/04/2018

∙

We analyze speed of convergence to global optimum for gradient descent training a deep linear neural network (parameterized as x W_N ... W_1x) by minimizing the ℓ_2 loss over whitened data. Convergence at a linear rate is guaranteed when the following hold: (i) dimensions of hidden layers are at least the minimum of the input and output dimensions; (ii) weight matrices at initialization are approximately balanced; and (iii) the initial loss is smaller than the loss of any rank-deficient solution. The assumptions on initialization (conditions (ii) and (iii)) are necessary, in the sense that violating any one of them may lead to convergence failure. Moreover, in the important case of output dimension 1, i.e. scalar regression, they are met, and thus convergence to global optimum holds, with constant probability under a random initialization scheme. Our results significantly extend previous analyses, e.g., of deep linear residual networks (Bartlett et al., 2018).

READ FULL TEXT

A Convergence Analysis of Gradient Descent for Deep Linear Neural Networks

On the Global Convergence of Training Deep Linear ResNets

Effects of Depth, Width, and Initialization: A Convergence Analysis of Layer-wise Training for Deep Linear Neural Networks

Directional Convergence Analysis under Spherically Symmetric Distribution

Revisiting "Qualitatively Characterizing Neural Network Optimization Problems"

Low-Rank plus Sparse Decomposition of Covariance Matrices using Neural Network Parametrization

Gradient descent with identity initialization efficiently learns positive definite linear transformations by deep residual networks

Global Convergence of Deep Networks with One Wide Layer Followed by Pyramidal Topology

A Convergence Analysis of Gradient Descent for Deep Linear Neural Networks

Related Research

On the Global Convergence of Training Deep Linear ResNets

Effects of Depth, Width, and Initialization: A Convergence Analysis of Layer-wise Training for Deep Linear Neural Networks

Directional Convergence Analysis under Spherically Symmetric Distribution

Revisiting "Qualitatively Characterizing Neural Network Optimization Problems"

Low-Rank plus Sparse Decomposition of Covariance Matrices using Neural Network Parametrization

Gradient descent with identity initialization efficiently learns positive definite linear transformations by deep residual networks

Global Convergence of Deep Networks with One Wide Layer Followed by Pyramidal Topology