Measuring Representational Robustness of Neural Networks Through Shared Invariances

TL;DR

This paper introduces a novel method called STIR to measure the robustness of neural networks by quantifying shared invariances between two models, moving beyond human-defined perturbations.

Contribution

It proposes a new measure, STIR, that assesses shared invariances between neural networks, generalizing robustness evaluation beyond human-centric perturbations.

Findings

Shared invariances vary with architecture and training methods.

STIR provides insights into neural network robustness.

The method is applicable to different network configurations.

Abstract

A major challenge in studying robustness in deep learning is defining the set of ``meaningless'' perturbations to which a given Neural Network (NN) should be invariant. Most work on robustness implicitly uses a human as the reference model to define such perturbations. Our work offers a new view on robustness by using another reference NN to define the set of perturbations a given NN should be invariant to, thus generalizing the reliance on a reference ``human NN'' to any NN. This makes measuring robustness equivalent to measuring the extent to which two NNs share invariances, for which we propose a measure called STIR. STIR re-purposes existing representation similarity measures to make them suitable for measuring shared invariances. Using our measure, we are able to gain insights into how shared invariances vary with changes in weight initialization, architecture, loss functions, and training dataset. Our implementation is available at: \url{https://github.com/nvedant07/STIR}.