A Fuzzy and Explainable AI Framework for Comparing Physical and Perceptual Representations in Galaxy Morphology

This record corresponds to the accepted manuscript (post-print) of the following journal article:

“A Fuzzy and Explainable AI Framework for Comparing Physical and Perceptual Representations in Galaxy Morphology”

This article introduces a fuzzy and explainable artificial intelligence framework designed to compare physically derived and human-based galaxy morphology classifications while explicitly accounting for uncertainty and interpretability. The framework combines photometric and structural features from the Sloan Digital Sky Survey (SDSS), Fuzzy C-Means clustering, Galaxy Zoo 2 debiased vote fractions, supervised learning models, and SHAP-based explainability to analyze the alignment between physical structure and human perception.

The methodology shows that physical clustering is mainly driven by structural concentration and bulge-dominance indicators, whereas human-based classifications exhibit smoother decision boundaries and greater sensitivity to photometric appearance. The study highlights the role of fuzzy logic and explainable AI in identifying transitional, ambiguous, and orientation-sensitive systems, providing a reproducible methodology for analyzing classification consistency, uncertainty, and human–model alignment in galaxy morphology.

The final published version is available at the publisher’s website:

https://doi.org/10.3390/ai7050159

This deposit is made for open access and dissemination purposes, in accordance with the publisher’s self-archiving policy.