Understanding How Over-Parametrization Leads to Acceleration: A case of learning a single teacher neuron

Abstract

Over-parametrization has become a popular technique in deep learning. It is observed that by over-parametrization, a larger neural network needs a fewer training iterations than a smaller one to achieve a certain level of performance -- namely, over-parametrization leads to acceleration in optimization. However, despite that over-parametrization is widely used nowadays, little theory is available to explain the acceleration due to over-parametrization. In this paper, we propose understanding it by studying a simple problem first. Specifically, we consider the setting that there is a single teacher neuron with quadratic activation, where over-parametrization is realized by having multiple student neurons learn the data generated from the teacher neuron. We provably show that over-parametrization helps the iterate generated by gradient descent to enter the neighborhood of a global optimal solution that achieves zero testing error faster. On the other hand, we also point out an issue regarding the necessity of over-parametrization and study how the scaling of the output neurons affects the convergence time.

arXiv commentAccepted at ACML 2021

References in corpus (19)

• Decoupled Weight Decay Regularization
• A Convergence Theory for Deep Learning via Over-Parameterization
• Train longer, generalize better: closing the generalization gap in large batch training of neural networks
• Implicit Regularization in Nonconvex Statistical Estimation: Gradient Descent Converges Linearly for Phase Retrieval, Matrix Completion, and Blind Deconvolution
• The loss surface of deep and wide neural networks
• On the Power of Over-parametrization in Neural Networks with Quadratic Activation
• On the Computational Efficiency of Training Neural Networks
• Implicit Regularization in Deep Matrix Factorization
• Efficient Learning of Generalized Linear and Single Index Models with Isotonic Regression
• Reliably Learning the ReLU in Polynomial Time
• Why do Larger Models Generalize Better? A Theoretical Perspective via the XOR Problem
• Optimization and Generalization of Shallow Neural Networks with Quadratic Activation Functions
• Time/Accuracy Tradeoffs for Learning a ReLU with respect to Gaussian Marginals
• Mildly Overparametrized Neural Nets can Memorize Training Data Efficiently
• Agnostic Learning of a Single Neuron with Gradient Descent
• Learning a Single Neuron with Gradient Methods
• Stationary Points of Shallow Neural Networks with Quadratic Activation Function
• Neural Networks and Polynomial Regression. Demystifying the Overparametrization Phenomena
• Avoiding Spurious Local Minima in Deep Quadratic Networks