iSAGE: An Incremental Version of SAGE for Online Explanation on Data Streams

TL;DR

iSAGE is an incremental, efficient extension of SAGE for online explanation of models in dynamic data environments, capable of handling concept drift and feature dependencies.

Contribution

We introduce iSAGE, a novel online explanation method that extends SAGE to handle streaming data, model updates, and feature dependency management.

Findings

iSAGE effectively detects and adapts to concept drift.

It maintains explanation accuracy in online settings.

Experimental results show improved efficiency over batch methods.

Abstract

Existing methods for explainable artificial intelligence (XAI), including popular feature importance measures such as SAGE, are mostly restricted to the batch learning scenario. However, machine learning is often applied in dynamic environments, where data arrives continuously and learning must be done in an online manner. Therefore, we propose iSAGE, a time- and memory-efficient incrementalization of SAGE, which is able to react to changes in the model as well as to drift in the data-generating process. We further provide efficient feature removal methods that break (interventional) and retain (observational) feature dependencies. Moreover, we formally analyze our explanation method to show that iSAGE adheres to similar theoretical properties as SAGE. Finally, we evaluate our approach in a thorough experimental analysis based on well-established data sets and data streams with concept drift.