Ilija Šimić, Eduardo Veas, Vedran Sabol. "A comprehensive analysis of perturbation methods in explainable AI feature attribution validation for neural time series classifiers." Scientific Reports, 2025. doi:10.1038/s41598-025-09538-2