AN ADAPTIVE HYBRID CLUSTERING MODEL USING FUZZY GUSTAFSON-KESSEL ALGORITHM WITH CHEETAH CHASE OPTIMIZATION FOR NONLINEAR DATA STRUCTURES

Clustering high-dimensional and anisotropic data remains a critical challenge in
unsupervised learning. While the Fuzzy Gustafson-Kessel (FGK) algorithm offers significant
improvements over traditional fuzzy clustering by accommodating elliptical clusters through
adaptive Mahalanobis distance, it is sensitive to initialization and prone to local optima. This
research proposes a novel hybrid model that integrates FGK with the Cheetah Chase
Algorithm (CCA), a nature-inspired metaheuristic optimization technique that mimics
predator-prey dynamics for global search efficiency. The CCA is employed to optimize the
initial cluster centers and fuzzifier parameters of FGK, enhancing its ability to adapt to
irregular, nonlinear data structures. The proposed FGK-CCA model iteratively refines the
evaluation on benchmark datasets and real-world applications in image segmentation and
bioinformatics demonstrates superior performance in terms of cluster validity indices,
convergence speed, and robustness against initialization. The FGK-CCA model represents a
powerful and adaptive solution for data clustering in dynamic and uncertain environments.