RobustX: Robust Counterfactual Explanations Made Easy

TL;DR

Robust explanations are crucial for trust in high-stakes ML decisions, but robustness of counterfactual explanations under changing scenarios remains challenging and lacks standardized benchmarks. The authors propose RobustX, an open-source Python library that implements a collection of CE generation and evaluation methods with a dedicated focus on robustness. RobustX offers interfaces to state-of-the-art CE methods, enabling streamlined access and facilitating benchmarking and comparison. The library is designed to be easily extensible, supporting rapid prototyping of novel robust CE generation and evaluation techniques, thereby advancing practical deployment and evaluation of robust explanations.

Abstract

The increasing use of Machine Learning (ML) models to aid decision-making in high-stakes industries demands explainability to facilitate trust. Counterfactual Explanations (CEs) are ideally suited for this, as they can offer insights into the predictions of an ML model by illustrating how changes in its input data may lead to different outcomes. However, for CEs to realise their explanatory potential, significant challenges remain in ensuring their robustness under slight changes in the scenario being explained. Despite the widespread recognition of CEs' robustness as a fundamental requirement, a lack of standardised tools and benchmarks hinders a comprehensive and effective comparison of robust CE generation methods. In this paper, we introduce RobustX, an open-source Python library implementing a collection of CE generation and evaluation methods, with a focus on the robustness property. RobustX provides interfaces to several existing methods from the literature, enabling streamlined access to state-of-the-art techniques. The library is also easily extensible, allowing fast prototyping of novel robust CE generation and evaluation methods.