On the Robustness of Convolutional Neural Networks to Internal Architecture and Weight Perturbations

03/23/2017
by   Nicholas Cheney, et al.
0

Deep convolutional neural networks are generally regarded as robust function approximators. So far, this intuition is based on perturbations to external stimuli such as the images to be classified. Here we explore the robustness of convolutional neural networks to perturbations to the internal weights and architecture of the network itself. We show that convolutional networks are surprisingly robust to a number of internal perturbations in the higher convolutional layers but the bottom convolutional layers are much more fragile. For instance, Alexnet shows less than a 30 performance when randomly removing over 70 convolutional or dense layers but performance is almost at chance with the same perturbation in the first convolutional layer. Finally, we suggest further investigations which could continue to inform the robustness of convolutional networks to internal perturbations.

READ FULL TEXT

page 1

page 2

page 3

page 4

research
12/22/2014

Deep Fried Convnets

The fully connected layers of a deep convolutional neural network typica...
research
02/25/2022

Understanding Adversarial Robustness from Feature Maps of Convolutional Layers

The adversarial robustness of a neural network mainly relies on two fact...
research
12/11/2019

Robust Gabor Networks

This work takes a step towards investigating the benefits of merging cla...
research
11/21/2018

Seeing in the dark with recurrent convolutional neural networks

Classical convolutional neural networks (cCNNs) are very good at categor...
research
11/18/2021

Wiggling Weights to Improve the Robustness of Classifiers

Robustness against unwanted perturbations is an important aspect of depl...
research
09/11/2018

On the Structural Sensitivity of Deep Convolutional Networks to the Directions of Fourier Basis Functions

Data-agnostic quasi-imperceptible perturbations on inputs can severely d...

Please sign up or login with your details

Forgot password? Click here to reset