# BASED-XAI: Breaking Ablation Studies Down for Explainable Artificial Intelligence

arXiv: https://arxiv.org/abs/2207.05566 · HTML: https://mlsanity.com/paper/2207.05566/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Builds on 11 of its 14 references that are in the map.
- First posted on arXiv 12 Jul 2022.

## Builds on (most cited first, 11 of 11)

- [A Unified Approach to Interpreting Model Predictions](https://mlsanity.com/paper/1705.07874/) (2017)
- [Axiomatic Attribution for Deep Networks](https://mlsanity.com/paper/1703.01365/) (2017)
- [Sanity Checks for Saliency Maps](https://mlsanity.com/paper/1810.03292/) (2018)
- [Captum: A unified and generic model interpretability library for PyTorch](https://mlsanity.com/paper/2009.07896/) (2020)
- [A Benchmark for Interpretability Methods in Deep Neural Networks](https://mlsanity.com/paper/1806.10758/) (2018)
- [Fooling LIME and SHAP: Adversarial Attacks on Post hoc Explanation Methods](https://mlsanity.com/paper/1911.02508/) (2019)
- [Explaining by Removing: A Unified Framework for Model Explanation](https://mlsanity.com/paper/2011.14878/) (2020)
- [Unrestricted Permutation forces Extrapolation: Variable Importance Requires at least One More Model, or There Is No Free Variable Importance](https://mlsanity.com/paper/1905.03151/) (2019)
- [The Out-of-Distribution Problem in Explainability and Search Methods for Feature Importance Explanations](https://mlsanity.com/paper/2106.00786/) (2021)
- [On Baselines for Local Feature Attributions](https://mlsanity.com/paper/2101.00905/) (2021)
- [Counterfactual Shapley Additive Explanations](https://mlsanity.com/paper/2110.14270/) (2021)

## Built on by (most cited first, 13 of 13)

- [BEExAI: Benchmark to Evaluate Explainable AI](https://mlsanity.com/paper/2407.19897/) (2024)
- [Generative AI voting: fair collective choice is resilient to LLM biases and inconsistencies](https://mlsanity.com/paper/2406.11871/) (2024)
- [The Disagreement Problem in Faithfulness Metrics](https://mlsanity.com/paper/2311.07763/) (2023)
- [Revisiting FunnyBirds evaluation framework for prototypical parts networks](https://mlsanity.com/paper/2408.11401/) (2024)
- [Upgrading Democracies with Fairer Voting Methods](https://mlsanity.com/paper/2505.14349/) (2025)
- [RUBEN: Rule-Based Explanations for Retrieval-Augmented LLM Systems](https://mlsanity.com/paper/2605.10862/) (2026)
- [Immersive Explainability: Visualizing Robot Navigation Decisions through XAI Semantic Scene Projections in Virtual Reality](https://mlsanity.com/paper/2504.00682/) (2025)
- [AICO: Feature Significance Tests for Supervised Learning](https://mlsanity.com/paper/2506.23396/) (2025)
- [Detection Transformers Under the Knife: A Neuroscience-Inspired Approach to Ablations](https://mlsanity.com/paper/2507.21723/) (2025)
- [Rule-Based Explanations for Retrieval-Augmented LLM Systems](https://mlsanity.com/paper/2510.22689/) (2025)
- [No Single Metric Tells the Whole Story: A Multi-Dimensional Evaluation Framework for Uncertainty Attributions](https://mlsanity.com/paper/2603.24524/) (2026)
- [A Personalized Computational Framework for Assessing the Sufficiency of Partially Observed Data in Healthcare AI models](https://mlsanity.com/paper/2607.09165/) (2026)
- [FADEx: Feature Attribution and Distortion-based Explanation of Dimensionality Reduction](https://mlsanity.com/paper/2607.27463/) (2026)

## Abstract (arXiv)

Explainable artificial intelligence (XAI) methods lack ground truth. In its place, method developers have relied on axioms to determine desirable properties for their explanations' behavior. For high stakes uses of machine learning that require explainability, it is not sufficient to rely on axioms as the implementation, or its usage, can fail to live up to the ideal. As a result, there exists active research on validating the performance of XAI methods. The need for validation is especially magnified in domains with a reliance on XAI. A procedure frequently used to assess their utility, and to some extent their fidelity, is an ablation study. By perturbing the input variables in rank order of importance, the goal is to assess the sensitivity of the model's performance. Perturbing important variables should correlate with larger decreases in measures of model capability than perturbing less important features. While the intent is clear, the actual implementation details have not been studied rigorously for tabular data. Using five datasets, three XAI methods, four baselines, and three perturbations, we aim to show 1) how varying perturbations and adding simple guardrails can help to avoid potentially flawed conclusions, 2) how treatment of categorical variables is an important consideration in both post-hoc explainability and ablation studies, and 3) how to identify useful baselines for XAI methods and viable perturbations for ablation studies.