Cluster-based method for the parameters evaluation and incomplete data studying in Hecht-Nielson neural network
This work describes the density-based static clustering algorithm for the efficient study of Hecht-Nielson neural network. Some original methods for noise patterns purging, missing values imputation in the training set, and random subsets bisection clustering algorithm for the Kohonen neuron parameters setting were applied.
Keywords: density-based clustering, missing data imputation, Hecht-Nielson neural network, bisection clustering
