Implicit Self-Regularization in Deep Neural Networks: Evidence from Random Matrix Theory and Implications for Learning

Abstract

Random Matrix Theory (RMT) is applied to analyze weight matrices of Deep Neural Networks (DNNs), including both production quality, pre-trained models such as AlexNet and Inception, and smaller models trained from scratch, such as LeNet5 and a miniature-AlexNet. Empirical and theoretical results clearly indicate that the DNN training process itself implicitly implements a form of Self-Regularization. The empirical spectral density (ESD) of DNN layer matrices displays signatures of traditionally-regularized statistical models, even in the absence of exogenously specifying traditional forms of explicit regularization. Building on relatively recent results in RMT, most notably its extension to Universality classes of Heavy-Tailed matrices, we develop a theory to identify 5+1 Phases of Training, corresponding to increasing amounts of Implicit Self-Regularization. These phases can be observed during the training process as well as in the final learned DNNs. For smaller and/or older DNNs, this Implicit Self-Regularization is like traditional Tikhonov regularization, in that there is a "size scale" separating signal from noise. For state-of-the-art DNNs, however, we identify a novel form of Heavy-Tailed Self-Regularization, similar to the self-organization seen in the statistical physics of disordered systems. This results from correlations arising at all size scales, which arises implicitly due to the training process itself. This implicit Self-Regularization can depend strongly on the many knobs of the training process. By exploiting the generalization gap phenomena, we demonstrate that we can cause a small model to exhibit all 5+1 phases of training simply by changing the batch size. This demonstrates that---all else being equal---DNN optimization with larger batch sizes leads to less-well implicitly-regularized models, and it provides an explanation for the generalization gap phenomena.

arXiv comment59 pages, 31 figures

Cited by in corpus (14)

• PowerSGD: Practical Low-Rank Gradient Compression for Distributed Optimization
• Data augmentation instead of explicit regularization
• Predicting trends in the quality of state-of-the-art neural networks without access to training or testing data
• Large batch size training of neural networks with adversarial training and second-order information
• The Full Spectrum of Deepnet Hessians at Scale: Dynamics with SGD Training and Sample Size
• Rethinking generalization requires revisiting old ideas: statistical mechanics approaches and complex learning behavior
• Memorization in Overparameterized Autoencoders
• Taxonomizing local versus global structure in neural network loss landscapes
• Parameter Re-Initialization through Cyclical Batch Size Schedules
• Deep Fundamental Factor Models
• Learning from learning machines: a new generation of AI technology to meet the needs of science
• Inefficiency of K-FAC for Large Batch Size Training
• Post-mortem on a deep learning contest: a Simpson's paradox and the complementary roles of scale metrics versus shape metrics
• Stable distributions and domains of attraction for unitarily invariant Hermitian random matrix ensembles