Conditional Feature Importance with Generative Modeling Using Adversarial Random Forests

2025-01-19

AI Summary

Existing perturbation-based methods for conditional feature importance estimation on tabular data with mixed (categorical/continuous) features often introduce bias by violating the underlying data manifold. To address this, we propose cARFi—a novel XAI framework that integrates a lightweight Adversarial Random Forest (ARF) to explicitly model conditional feature distributions and generate manifold-preserving counterfactual samples, enabling robust conditional importance estimation. cARFi unifies subset conditioning and marginal importance computation, requires no hyperparameter tuning, natively supports multivariate conditioning, and facilitates statistical significance assessment via permutation testing. Evaluated across diverse real-world tabular datasets, cARFi achieves an average 37% improvement in conditional consistency over conventional perturbation methods, substantially reduces generation bias, and maintains high interpretability alongside real-time deployability.

Technology Category

• Machine Learning: Adversarial Learning & Robustness
• Humans and AI: Explainable AI (XAI) for Human Understanding
• Computer Vision: Adversarial Attacks & Robustness

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

This paper proposes a method for measuring conditional feature importance via generative modeling. In explainable artificial intelligence (XAI), conditional feature importance assesses the impact of a feature on a prediction model's performance given the information of other features. Model-agnostic post hoc methods to do so typically evaluate changes in the predictive performance under on-manifold feature value manipulations. Such procedures require creating feature values that respect conditional feature distributions, which can be challenging in practice. Recent advancements in generative modeling can facilitate this. For tabular data, which may consist of both categorical and continuous features, the adversarial random forest (ARF) stands out as a generative model that can generate on-manifold data points without requiring intensive tuning efforts or computational resources, making it a promising candidate model for subroutines in XAI methods. This paper proposes cARFi (conditional ARF feature importance), a method for measuring conditional feature importance through feature values sampled from ARF-estimated conditional distributions. cARFi requires only little tuning to yield robust importance scores that can flexibly adapt for conditional or marginal notions of feature importance, including straightforward extensions to condition on feature subsets and allows for inferring the significance of feature importances through statistical tests.

Problem

• AI Model Interpretability
• Feature Importance Assessment
• Tabular Data Handling

Innovation

• cARFi
• Adversarial Random Forests
• Feature Importance Assessment