Symmetry-Preserving Paths in Integrated Gradients
Miguel Lerma, Mirtha Lucas
TL;DR
This paper proves that Integrated Gradients, a method for attributing importance in deep networks, satisfies key properties like completeness and symmetry preservation, and explores its uniqueness among path methods.
Contribution
It provides rigorous proofs of IG's properties and investigates its uniqueness as a symmetry-preserving path method.
Findings
IG satisfies completeness and symmetry-preserving properties
IG is unique among symmetry-preserving path methods
Theoretical validation of IG's foundational properties
Abstract
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.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsSparse and Compressive Sensing Techniques · Topological and Geometric Data Analysis · Stochastic Gradient Optimization Techniques