ExplainabilityTransparencyTrustworthinesscs.LG
A New Technique for AI Explainability using Feature Association Map
Sayantani Ghosh, Amit Kumar Das, Amlan Chakrabarti
May 12, 2026
arXiv.org
Key Takeaways
-
AI transparency requires better explanation tools: The paper addresses a critical gap in responsible AI by noting that lack of transparency in AI systems creates challenges for real-world applications, particularly in high-stakes domains where understanding why an AI made a decision is essential for building trust.
-
A new explanation method outperforms existing approaches: FAMeX, the proposed algorithm, demonstrated superior performance compared to established explainability methods like SHAP and Permutation Feature Importance across eight benchmark algorithms, suggesting it may provide clearer insights into which features drive AI decisions.
-
Feature relationships matter for explanations: Unlike simpler approaches, FAMeX uses a graph-based model that captures how features associate with and influence each other, potentially offering more nuanced explanations of AI decision-making than methods that treat features in isolation.
-
Better explainability supports responsible AI deployment: By improving how well we can explain AI predictions, tools like FAMeX help organizations meet accountability requirements and enable stakeholders—including regulators, users, and affected individuals—to understand and scrutinize AI system behavior in critical applications.
Abstract
Lack of transparency in AI systems poses challenges in critical real-life applications. It is important to be able to explain the decisions of an AI system to ensure trust on the system. Explainable AI (XAI) algorithms play a vital role in achieving this objective. In this paper, we are proposing a new algorithm for Explaining AI systems, FAMeX (Feature Association Map based eXplainability). The proposed algorithm is based on a graph-theoretic formulation of the feature set termed as Feature Association Map (FAM). The foundation of the modelling is based on association between features. The proposed FAMeX algorithm has been found to be better than the competing XAI algorithms - Permutation Feature Importance (PFI) and SHapley Additive exPlanations (SHAP). Experiments conducted with eight benchmark algorithms show that FAMeX is able to gauge feature importance in the context of classification better than the competing algorithms. This definitely shows that FAMeX is a promising algorithm in explaining the predictions from an AI system