Robustness to Pruning Predicts Generalization in Deep Neural Networks

by   Lorenz Kuhn, et al.

Existing generalization measures that aim to capture a model's simplicity based on parameter counts or norms fail to explain generalization in overparameterized deep neural networks. In this paper, we introduce a new, theoretically motivated measure of a network's simplicity which we call prunability: the smallest fraction of the network's parameters that can be kept while pruning without adversely affecting its training loss. We show that this measure is highly predictive of a model's generalization performance across a large set of convolutional networks trained on CIFAR-10, does not grow with network size unlike existing pruning-based measures, and exhibits high correlation with test set loss even in a particularly challenging double descent setting. Lastly, we show that the success of prunability cannot be explained by its relation to known complexity measures based on models' margin, flatness of minima and optimization speed, finding that our new measure is similar to – but more predictive than – existing flatness-based measures, and that its predictions exhibit low mutual information with those of other baselines.


page 1

page 2

page 3

page 4


A Reparameterization-Invariant Flatness Measure for Deep Neural Networks

The performance of deep neural networks is often attributed to their aut...

Feature-Robustness, Flatness and Generalization Error for Deep Neural Networks

The performance of deep neural networks is often attributed to their aut...

The Generalization-Stability Tradeoff in Neural Network Pruning

Pruning neural network parameters to reduce model size is an area of muc...

The intriguing role of module criticality in the generalization of deep networks

We study the phenomenon that some modules of deep neural networks (DNNs)...

Why Flatness Correlates With Generalization For Deep Neural Networks

The intuition that local flatness of the loss landscape is correlated wi...

Fantastic Generalization Measures and Where to Find Them

Generalization of deep networks has been of great interest in recent yea...

Are All Linear Regions Created Equal?

The number of linear regions has been studied as a proxy of complexity f...

Please sign up or login with your details

Forgot password? Click here to reset