Accelerated learning algorithms of general fuzzy min-max neural network using a novel hyperbox selection rule

• Proposing a novel theoretical solution to accelerate learning algorithms of GFMM. Providing proofs mathematical lemmas for the proposed method. Assessing efficiency method on widely used datasets. The training time in GFMM model is significantly reduced. This paper proposes process general fuzzy min-max neural network. purpose reduce unsuitable hyperboxes selected as potential candidates expansion step existing cover new input pattern online or hyperbox aggregation agglomerative algorithms. Our approach based formulas form aiming remove which are certain not satisfy conditions, and turn decreasing assessed over number data sets. experimental results indicated significant decrease both Notably, reduced from 1.2 12 times when using method, while accelerated 7 37 average.