Learning Model-Agnostic Counterfactual Explanations for Tabular Data

Abstract

Counterfactual explanations can be obtained by identifying the smallest change made to a feature vector to qualitatively influence a prediction; for example, from 'loan rejected' to 'awarded' or from 'high risk of cardiovascular disease' to 'low risk'. Previous approaches often emphasized that counterfactuals should be easily interpretable to humans, motivating sparse solutions with few changes to the feature vectors. However, these approaches would not ensure that the produced counterfactuals be proximate (i.e., not local outliers) and connected to regions with substantial data density (i.e., close to correctly classified observations), two requirements known as counterfactual faithfulness. These requirements are fundamental when making suggestions to individuals that are indeed attainable. Our contribution is twofold. On one hand, we suggest to complement the catalogue of counterfactual quality measures [1] using a criterion to quantify the degree of difficulty for a certain counterfactual suggestion. On the other hand, drawing ideas from the manifold learning literature, we develop a framework that generates attainable counterfactuals. We suggest the counterfactual conditional heterogeneous variational autoencoder (C-CHVAE) to identify attainable counterfactuals that lie within regions of high data density.

arXiv commentUpdate version: from Neurips Workshop to WWW publication. In Proceedings of The Web Conference 2020 (WWW 20), April 20-24, 2020, Taipei, Taiwan

References in corpus (13)

• Auto-Encoding Variational Bayes
• Fairness Beyond Disparate Treatment & Disparate Impact: Learning Classification without Disparate Mistreatment
• Actionable Recourse in Linear Classification
• Wasserstein Auto-Encoders
• Adversarial Patch
• Interpretable Predictions of Tree-based Ensembles via Actionable Feature Tweaking
• Towards Realistic Individual Recourse and Actionable Explanations in Black-Box Decision Making Systems
• Interpretable Credit Application Predictions With Counterfactual Explanations
• Human Perceptions of Fairness in Algorithmic Decision Making: A Case Study of Criminal Risk Prediction
• Variational Autoencoder with Arbitrary Conditioning
• Generalized Inverse Classification
• Inverse Classification for Comparison-based Interpretability in Machine Learning
• Issues with post-hoc counterfactual explanations: a discussion

Cited by in corpus (41)

• Deep Neural Networks and Tabular Data: A Survey
• From Anecdotal Evidence to Quantitative Evaluation Methods: A Systematic Review on Evaluating Explainable AI
• A Survey on XAI for 5G and Beyond Security: Technical Aspects, Challenges and Research Directions
• Counterfactual Explanations for Machine Learning on Multivariate Time Series Data
• Counterfactual Shapley Additive Explanations
• A survey of algorithmic recourse: definitions, formulations, solutions, and prospects
• Model-Based Counterfactual Synthesizer for Interpretation
• ReLAX: Reinforcement Learning Agent eXplainer for Arbitrary Predictive Models
• Beyond Individualized Recourse: Interpretable and Interactive Summaries of Actionable Recourses
• Counterfactuals and Causability in Explainable Artificial Intelligence: Theory, Algorithms, and Applications
• Benchmarking Instance-Centric Counterfactual Algorithms for XAI: From White Box to Black Box
• Achieving Diversity in Counterfactual Explanations: a Review and Discussion
• Latent-CF: A Simple Baseline for Reverse Counterfactual Explanations
• Interpretable Machine Learning: Moving From Mythos to Diagnostics
• CARLA: A Python Library to Benchmark Algorithmic Recourse and Counterfactual Explanation Algorithms
• Counterfactual Explanations for Arbitrary Regression Models
• Feature-based Learning for Diverse and Privacy-Preserving Counterfactual Explanations
• TABCF: Counterfactual Explanations for Tabular Data Using a Transformer-Based VAE
• Algorithmic Recourse in the Wild: Understanding the Impact of Data and Model Shifts
• Consistent Counterfactuals for Deep Models
• RoCourseNet: Distributionally Robust Training of a Prediction Aware Recourse Model
• A Survey on the Robustness of Feature Importance and Counterfactual Explanations
• The Intriguing Relation Between Counterfactual Explanations and Adversarial Examples
• Ordered Counterfactual Explanation by Mixed-Integer Linear Optimization
• Evaluating Model Explanations without Ground Truth
• A unified framework for evaluating the robustness of machine-learning interpretability for prospect risking
• Exploring Counterfactual Explanations Through the Lens of Adversarial Examples: A Theoretical and Empirical Analysis
• Framing Algorithmic Recourse for Anomaly Detection
• On Quantitative Evaluations of Counterfactuals
• Flexible and Robust Counterfactual Explanations with Minimal Satisfiable Perturbations
• When Can You Trust Your Explanations? A Robustness Analysis on Feature Importances
• Towards User Guided Actionable Recourse
• Navigating Explanatory Multiverse Through Counterfactual Path Geometry
• MCCE: Monte Carlo sampling of realistic counterfactual explanations
• CounterNet: End-to-End Training of Prediction Aware Counterfactual Explanations
• Perfect Counterfactuals in Imperfect Worlds: Modelling Noisy Implementation of Actions in Sequential Algorithmic Recourse
• Adaptive Adversarial Training Does Not Increase Recourse Costs
• Hierarchical Variational Autoencoder for Visual Counterfactuals
• Realistic Counterfactual Explanations via Denial Constraints
• Counterfactual Explanations via Latent Space Projection and Interpolation
• Low-Cost Algorithmic Recourse for Users With Uncertain Cost Functions