AUTHOR=Björklund Anton , Henelius Andreas , Oikarinen Emilia , Kallonen Kimmo , Puolamäki Kai 

TITLE=Explaining any black box model using real data

JOURNAL=Frontiers in Computer Science

VOLUME=Volume 5 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/computer-science/articles/10.3389/fcomp.2023.1143904

DOI=10.3389/fcomp.2023.1143904

ISSN=2624-9898

ABSTRACT=In recent years the use of complex machine learning has increased drastically. However, these complex black box models trade interpretability for accuracy, which is not always acceptable. It is particularly troubling in, e.g., socially sensitive, safety-critical, or knowledge extraction applications. The use of non-interpretable models has led to the development of various explanation methods for interpreting predictions from black box models.

In this paper, we propose a new explanation method, SLISE, which can be used with any black box model (model-agnostic), does not require any modifications to the black box model (post-hoc),  and explains individual predictions (local). Our approach solves shortcomings in other related explanation methods by only using existing data instead of sampling new, artificial data. The method is also usable without modifications across various data domains. We demonstrate and evaluate our method using real-world datasets and compare it against other state-of-the-art model-agnostic, local explanation methods.