Exact solutions to the nonlinear dynamics of learning in deep linear neural networks

by   Andrew M. Saxe, et al.

Despite the widespread practical success of deep learning methods, our theoretical understanding of the dynamics of learning in deep neural networks remains quite sparse. We attempt to bridge the gap between the theory and practice of deep learning by systematically analyzing learning dynamics for the restricted case of deep linear neural networks. Despite the linearity of their input-output map, such networks have nonlinear gradient descent dynamics on weights that change with the addition of each new hidden layer. We show that deep linear networks exhibit nonlinear learning phenomena similar to those seen in simulations of nonlinear networks, including long plateaus followed by rapid transitions to lower error solutions, and faster convergence from greedy unsupervised pretraining initial conditions than from random initial conditions. We provide an analytical description of these phenomena by finding new exact solutions to the nonlinear dynamics of deep learning. Our theoretical analysis also reveals the surprising finding that as the depth of a network approaches infinity, learning speed can nevertheless remain finite: for a special class of initial conditions on the weights, very deep networks incur only a finite, depth independent, delay in learning speed relative to shallow networks. We show that, under certain conditions on the training data, unsupervised pretraining can find this special class of initial conditions, while scaled random Gaussian initializations cannot. We further exhibit a new class of random orthogonal initial conditions on weights that, like unsupervised pre-training, enjoys depth independent learning times. We further show that these initial conditions also lead to faithful propagation of gradients even in deep nonlinear networks, as long as they operate in a special regime known as the edge of chaos.


Statistical Mechanics of Deep Linear Neural Networks: The Back-Propagating Renormalization Group

The success of deep learning in many real-world tasks has triggered an e...

Gradient flow dynamics of shallow ReLU networks for square loss and orthogonal inputs

The training of neural networks by gradient descent methods is a corners...

On the Neural Tangent Kernel of Deep Networks with Orthogonal Initialization

In recent years, a critical initialization scheme with orthogonal initia...

A mathematical theory of semantic development in deep neural networks

An extensive body of empirical research has revealed remarkable regulari...

On the Learning Dynamics of Deep Neural Networks

While a lot of progress has been made in recent years, the dynamics of l...

Simple lessons from complex learning: what a neural network model learns about cosmic structure formation

We train a neural network model to predict the full phase space evolutio...

Provable Benefit of Orthogonal Initialization in Optimizing Deep Linear Networks

The selection of initial parameter values for gradient-based optimizatio...

Please sign up or login with your details

Forgot password? Click here to reset