Answers are grounded in this paper and cite their sources.
In this paper, we investigate the relationship between uncertainty and information from the perspectives of information theory and fuzzy theory. We propose a new three-parameter generalized fuzzy entropy measure that unifies several existing fuzzy entropy measures, including those of De Luca-Termini and Hooda, as special cases. We prove that the proposed measure satisfies all four required properties of a fuzzy entropy measure and establish its monotonicity with respect to each parameter for certain values of β . We further analyze the concavity, convexity, and log-concavity properties of the kernel function, which, together with the concavity of the kernel function, imply the subadditivity of the proposed measure under algebraic operations when a certain concavity condition on the parameters is satisfied. Asymptotic properties of the probabilistic form are also derived, showing convergence to Shannon entropy and the max-entropy principle. Moreover, we characterize the maximizer distribution under a mean constraint, derive the closed-form partial derivative for gradient-based optimization, prove invariance under generalized complementary transformations, and demonstrate that increasing one parameter reduces noise sensitivity, making the measure suitable for feature selection. Numerical comparisons evaluate the performance of the proposed measure against existing measures across multiple fuzziness scenarios.We further validate the practical utility of the proposed measure through a real-world image segmentation application using the benchmark Lena image, including gradient-based threshold optimization, noise robustness analysis, histogram equalization validation, dual invariance verification, and computational time evaluation. Additionally, we provide complementary validation on tabular data from the UCI Machine Learning Repository to demonstrate the generality of the proposed measure beyond image processing applications.
No takes yet. Share an insight, caveat, or question.
Mehdi Shams (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: