GLEAMS: Bridging the Gap Between Local and Global Explanations

TL;DR

GLEAMS is a new explainability method that combines local and global explanations by partitioning input space and learning interpretable models within each region, offering more faithful and understandable insights.

Contribution

The paper introduces GLEAMS, a novel approach that bridges local and global explanation gaps by partitioning input space and learning interpretable models in each segment.

Findings

Effective on synthetic and real-world data

Provides faithful local and global explanations

Yields human-understandable insights

Abstract

The explainability of machine learning algorithms is crucial, and numerous methods have emerged recently. Local, post-hoc methods assign an attribution score to each feature, indicating its importance for the prediction. However, these methods require recalculating explanations for each example. On the other side, while there exist global approaches they often produce explanations that are either overly simplistic and unreliable or excessively complex. To bridge this gap, we propose GLEAMS, a novel method that partitions the input space and learns an interpretable model within each sub-region, thereby providing both faithful local and global surrogates. We demonstrate GLEAMS' effectiveness on both synthetic and real-world data, highlighting its desirable properties and human-understandable insights.