Generalized Integrated Gradients: A practical method for explaining diverse ensembles

By J. Merrill, Geoff Ward, S. Kamkar, Jay Budzik, Douglas C. Merrill. 2019

16 citations recorded by the source corpus.

We introduce Generalized Integrated Gradients (GIG), a formal extension of the Integrated Gradients (IG) (Sundararajan et al., 2017) method for attributing credit to the input variables of a predictive model. GIG improves IG by explaining a broader variety of functions that arise from practical applications of ML in domains like financial services. GIG is constructed to overcome limitations of Shapley (1953) and Aumann-Shapley (1974), and has desirable properties when compared to other approaches. We prove GIG is the only correct method, under a small set of reasonable axioms, for providing explanations for mixed-type models or games. We describe the implementation, and present results of experiments on several datasets and systems of models.

Subfields: Artificial Intelligence, Information Systems and Management

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Su-In Lee
2. Scott M. Lundberg
3. Hugh Chen
4. Przemyslaw Biecek
5. Adrian Weller
6. David Watson
7. Ankur Taly
8. Mukund Sundararajan

In Information Systems and Management

1. Su-In Lee
2. Scott M. Lundberg

Citations in the corpus

• Explaining Models by Propagating Shapley Values of Local Components (2019, background): “…numerous works: Occlusion (Zeiler and Fergus 2014), Saliency Maps (Simonyan, Vedaldi, and Zisserman 2013), Layer-Wise Relevance Propagation (Bach et al. 2015), DeepLIFT, Integrated Gradients (IG) (Sundararajan, Taly, and Yan 2017), and Generalized Integrated Gradients (GIG) (Merrill et al. 2019).”