Explaining the Explainer: A First Theoretical Analysis of LIME

Damien Garreau (1, 2, 3) , Ulrike von Luxburg (1, 4)

1 Max Planck Institute for Intelligent Systems [Tübingen]
2 LJAD - Laboratoire Jean Alexandre Dieudonné
3 MAASAI - Modèles et algorithmes pour l’intelligence artificielle
4 University of Tübingen

Machine learning is used more and more often for sensitive applications, sometimes replacing humans in critical decision-making processes. As such, interpretability of these algorithms is a pressing need. One popular algorithm to provide interpretability is LIME (Local Interpretable Model-Agnostic Explanation). In this paper, we provide the first theoretical analysis of LIME. We derive closed-form expressions for the coefficients of the interpretable model when the function to explain is linear. The good news is that these coefficients are proportional to the gradient of the function to explain: LIME indeed discovers meaningful features. However, our analysis also reveals that poor choices of parameters can lead LIME to miss important features.

Domaines: Intelligence artificielle [cs.AI]; Machine Learning [stat.ML]

Damien Garreau, Ulrike von Luxburg. Explaining the Explainer: A First Theoretical Analysis of LIME. AISTATS 2020 - 23rd International Conference on Artificial Intelligence and Statistics, Aug 2020, Palermo /Online, Italy. ⟨hal-03233013v2⟩