A faster dynamic convergency approach for self-organizing maps

Abstract This paper proposes a novel variable learning rate to address two main challenges of the conventional Self-Organizing Maps (SOM) termed VLRSOM: high accuracy with fast convergence and low topological error. We empirically showed that proposed method exhibits faster behavior. It is also more robust in topology preservation as it maintains an optimal until end maximum iterations. Since adaption misadjustment parameter depends on calculated error, VLRSOM will avoid undesired results by exploiting error response during weight updation. Then updated adaptively after random initialization at beginning training process. Experimental show eliminates tradeoff between data's relationship. Extensive experiments were conducted different types datasets evaluate performance method. First, we experimented synthetic data handwritten digits. For each set, number iterations (200 500) performed test stability network. The was further evaluated using four benchmark sets. These include Balance, Wisconsin Breast, Dermatology, Ionosphere. In addition, comprehensive comparative analysis three other SOM techniques: SOM, parameter-less self-organizing map (PLSOM2), RA-SOM terms accuracy, quantization (QE), (TE). indicated approach produced superior methods.