id	author	title	date	pages	extension	mime	words	sentence	flesch	summary	cache	txt
cana-4650	Mintu Debnath	Mathematical Foundations of Explainable AI: A Framework based on Topological Data Analysis	2025	24	.pdf	application/pdf	5549	276	38	• Iris features are moderately skewed with light-tailed distributions. B. Explanation Configuration Parameters SHAP Settings • Explainer: KernelExplainer (for LR, SVM, RF), DeepExplainer (for NN) • Background samples: 100 • Perturbation samples: 500 • Link function: Identity • Summarization method: k-means LIME Settings • Perturbations per sample: 500 • Kernel width: 0.75 • Local surrogate: Ridge regression (L2) • Feature selection: Forward selection Permutation Feature Importance Settings • Number of repeats: 10 Communications on Applied Nonlinear Analysis ISSN: 1074-133X Vol 32 No. 9s (2025) 3140 https://internationalpubls.com • Scoring metric: AUC (binary), Macro-F1 (multi-class) • Random seed: 42 C. Interpretability Metrics: Definitions and Implementation 4.	cache/cana-4650.pdf	txt/cana-4650.txt