Improved online learning algorithm combined with K-nearest hyperbox expansion for general fuzzy min–max neural networks

Abstract An improved classifier based on an online learning algorithm for the general fuzzy min–max neural network (IOL_GFMM), integrated with a k-nearest hyperbox expansion rule, is proposed in this paper. The primary motivation is to significantly reduce training time while preserving, and in some cases improving, classification performance. Unlike the conventional IOL_GFMM, which evaluates all hyperboxes belonging to the same class during the expansion process, the proposed method restricts the evaluation to only the k-nearest hyperboxes, thereby reducing computational complexity. This study represents the first attempt to integrate IOL_GFMM, a contraction-free fuzzy min–max neural network, with a k-nearest hyperbox expansion strategy, referred to as IOL_GFMM_KNN. The performance of the proposed approach is evaluated using multiple benchmark datasets and assessed based on performance metrics, including classification accuracy and training time. Experimental results demonstrate that the proposed method achieves comparable or improved classification accuracy while providing a substantial reduction in computational time compared to the original IOL_GFMM. The proposed approach has the potential to offer significant societal benefits by enabling the development of faster and more reliable decision support systems for real-world applications such as medical diagnosis, bioinformatics, and large-scale data-driven classification problems.