Concerning Uncertainty -- A Systematic Survey of Uncertainty-Aware XAI

1. 1Identify uncertainty quantification methods (Bayesian, Monte Carlo, Conformal)
2. 2Integrate uncertainty into explanations using strategies such as assessing trustworthiness and constraining models
3. 3Evaluate UAXAI methods using appropriate metrics and benchmarks
4. 4Apply UAXAI to real-world problems to improve model reliability and transparency

AI engineers and researchers benefit from this survey as it provides a comprehensive overview of uncertainty quantification methods and their applications in explainable AI, enabling them to develop more robust and trustworthy models

💡 Incorporating uncertainty into explainable AI pipelines is crucial for developing trustworthy models

🔍 Uncertainty-aware XAI survey: Bayesian, Monte Carlo, and Conformal methods for robust explanations