A Dynamical Central Limit Theorem for Shallow Neural Networks

Abstract

Recent theoretical works have characterized the dynamics of wide shallow neural networks trained via gradient descent in an asymptotic mean-field limit when the width tends towards infinity. At initialization, the random sampling of the parameters leads to deviations from the mean-field limit dictated by the classical Central Limit Theorem (CLT). However, since gradient descent induces correlations among the parameters, it is of interest to analyze how these fluctuations evolve. Here, we use a dynamical CLT to prove that the asymptotic fluctuations around the mean limit remain bounded in mean square throughout training. The upper bound is given by a Monte-Carlo resampling error, with a variance that that depends on the 2-norm of the underlying measure, which also controls the generalization error. This motivates the use of this 2-norm as a regularization term during training. Furthermore, if the mean-field dynamics converges to a measure that interpolates the training data, we prove that the asymptotic deviation eventually vanishes in the CLT scaling. We also complement these results with numerical experiments.

Appeared in Advances in Neural Information Processing Systems 33 (NeurIPS 2020). An error in Theorem 3.5 has been corrected

References in corpus (6)

• Deep Neural Networks as Gaussian Processes
• Fine-Grained Analysis of Optimization and Generalization for Overparameterized Two-Layer Neural Networks
• The Gaussian equivalence of generative models for learning with shallow neural networks
• Asymptotics of Wide Networks from Feynman Diagrams
• On the Banach spaces associated with multi-layer ReLU networks: Function representation, approximation theory and gradient descent dynamics
• On Sparsity in Overparametrised Shallow ReLU Networks

Cited by in corpus (1)

• On the existence of global minima and convergence analyses for gradient descent methods in the training of deep neural networks