Hardware Implementation Issues of the Neighborhood Mechanism in Kohonen Self Organized Feature Maps Marta Kolasa 1 and Rafal Dlugosz 2,* 1- University of Technology and Life Sciences, Institute of Electrical Engineering, ul. Kaliskiego 7, 85-791, Bydgoszcz, Poland 2- Swiss Federal Institute of Technology in Lausanne, Institute of Microtechnology, Rue A.-L. Breguet 2, CH-2000, Neuchâtel, Switzerland Abstract. In this paper, we discuss an important problem of the selection of the neighborhood radius in the learning schemes of the Winner Takes Most Kohonen neural network. The optimization of this parameter is essential in case of hardware realization of the network given that the lower values of the radius can result in significant reduction of both the power dissipation and the chip area, even by 40-60% that is important in application of such networks in low power devices. The simulation studies reveal that using large initial values of the neighborhood radius usually is not the most optimal. For a wide range of the training parameters some optimal values, usually small, of the neighborhood radius may be indicated that allow for the minimization of the quantization error. 1 Introduction Kohonen neural networks (KNN), often referred to as self organized feature maps (SOFM), belong to the group of the networks that are trained without supervision. They usually consist of a single layer of neurons that are organized in a map-like structure with different grids. KNNs have been broadly described in the literature [1, 2], along with various optimization techniques of the learning algorithm [3, 4]. KNNs usually are realized using software platform, however many attempts to realize them on transistor level were undertaken in the past [5]. Hardware implement- tations create some specific problems that are of second importance in the software realizations, e.g. the necessity of optimization of the energy consumption and the chip area. In this paper we show that the key parameters of the learning algorithm e.g. the neighborhood radius have a great impact on these parameters and need optimization. The paper is organized as follows. In next section we present shortly the idea of SOFM. Next we present the hardware implementation issues of such network. Then we analyze simulation results. The conclusions are formulated at the end. 2 Kohonen Neural Network The competitive learning in Kohonen networks is an iterative process, in which all training patterns of a given learning set are in particular iterations presented to the * this work is supported by EU Marie Curie OIF fellowship, project No. 21926 ESANN'2009 proceedings, European Symposium on Artificial Neural Networks - Advances in Computational Intelligence and Learning. Bruges (Belgium), 22-24 April 2009, d-side publi., ISBN 2-930307-09-9. 565