Symmetry-Preserving Paths in Integrated Gradients

03/25/2021

∙

We provide rigorous proofs that the Integrated Gradients (IG) attribution method for deep networks satisfies completeness and symmetry-preserving properties. We also study the uniqueness of IG as a path method preserving symmetry.

READ FULL TEXT

Symmetry-Preserving Paths in Integrated Gradients

Investigating Saturation Effects in Integrated Gradients

An exploration of the influence of path choice in game-theoretic attribution algorithms

Compensated Integrated Gradients to Reliably Interpret EEG Classification

Attribution in Scale and Space

Negentropic Planar Symmetry Detector

A Rigorous Study of Integrated Gradients Method and Extensions to Internal Neuron Attributions

Extended Symmetry Preserving Attention Networks for LHC Analysis

Symmetry-Preserving Paths in Integrated Gradients

Related Research

Investigating Saturation Effects in Integrated Gradients

An exploration of the influence of path choice in game-theoretic attribution algorithms

Compensated Integrated Gradients to Reliably Interpret EEG Classification

Attribution in Scale and Space

Negentropic Planar Symmetry Detector

A Rigorous Study of Integrated Gradients Method and Extensions to Internal Neuron Attributions

Extended Symmetry Preserving Attention Networks for LHC Analysis