Increasing Performance And Sample Efficiency With Model-agnostic Interactive Feature Attributions

TL;DR

This paper introduces a model-agnostic interactive explanation method that corrects feature attributions to improve model performance and sample efficiency through data augmentation and expert input.

Contribution

It provides implementations for interactive correction of feature attributions using Occlusion and Shapley values, enhancing model training and active learning.

Findings

Significant performance improvements via attribution-based data augmentation.

Enhanced sample efficiency in active learning with interactive explanations.

Effective correction of feature attributions using domain expert input.

Abstract

Model-agnostic feature attributions can provide local insights in complex ML models. If the explanation is correct, a domain expert can validate and trust the model's decision. However, if it contradicts the expert's knowledge, related work only corrects irrelevant features to improve the model. To allow for unlimited interaction, in this paper we provide model-agnostic implementations for two popular explanation methods (Occlusion and Shapley values) to enforce entirely different attributions in the complex model. For a particular set of samples, we use the corrected feature attributions to generate extra local data, which is used to retrain the model to have the right explanation for the samples. Through simulated and real data experiments on a variety of models we show how our proposed approach can significantly improve the model's performance only by augmenting its training dataset based on corrected explanations. Adding our interactive explanations to active learning settings increases the sample efficiency significantly and outperforms existing explanatory interactive strategies. Additionally we explore how a domain expert can provide feature attributions which are sufficiently correct to improve the model.