Explaining a prediction in some nonlinear models

TL;DR

This paper introduces a method combining integrated gradient and deep Taylor decomposition to explain input contributions in nonlinear models, including neural networks, with a natural reference point selection.

Contribution

It presents a novel approach that merges existing explanation methods, offering a natural reference point tailored to the specific model used.

Findings

Applicable to regression and classification models

Provides a natural reference point for explanations

Enhances interpretability of deep neural networks

Abstract

In this article we will analyse how to compute the contribution of each input value to its aggregate output in some nonlinear models. Regression and classification applications, together with related algorithms for deep neural networks are presented. The proposed approach merges two methods currently present in the literature: integrated gradient and deep Taylor decomposition. Compared to DeepLIFT and Deep SHAP, it provides a natural choice of the reference point peculiar to the model at use.