Single Dendritic Neuron Model Trained by Spherical Search Algorithm For Classification

Recently, the single dendritic neuron with non-linear locality has been focused by many neuroscience researches. The dendrites are composed of several branches, and these branches correspond to three distributions in coordinate, which are used to classify the training data as required. To develop a new dendritic neuron model (DNM) for solving more practical problems, we use a recently proposed state-of-the-art optimization algorithm called spherical search algorithm as the training algorithm. The results based on several real-world classification tasks suggest that the proposed learning algorithm is more effective and promising for training DNM in comparison with another recently proposed differential evolution learning algorithm. It thereafter indicates that DNM, as a single neuron model, is a suitable and powerful classifier.