A similarity measure for intuitionistic fuzzy sets with improved clustering accuracy and applications in pattern recognition

Modern systems often encounter uncertainties arising from both randomness and fuzziness, which pose significant challenges to effective data analysis. To address these challenges, researchers have increasingly focused on fuzzy clustering techniques as a unified approach to managing dual uncertainties. This manuscript proposes a modified variant of the intuitionistic fuzzy c-means clustering technique coupled with an improved genetic algorithm to refine clustering precision. The proposed method mitigates noise sensitivity by developing a robust similarity metric, while Yager’s negation function effectively handles uncertainty in membership value assignments. Additionally, the improved genetic algorithm employs normalized crossover and mutation operators to mitigate sensitivity to initial centroid selection. The method demonstrates superior adaptability compared to existing approaches, effectively addressing key clustering challenges. Its efficacy is validated through experiments on benchmark datasets, with comparative analyses against existing state-of-the-art techniques by delivering higher accuracy, maximizing inter-cluster distance, reducing objective function value and minimizing sensitivity analysis. These findings demonstrate that the proposed IGAAIFCM technique successfully addresses key limitations of conventional fuzzy clustering by delivering a robust and uncertainty-adaptive clustering framework with improved stability and accuracy.