A Neuron Model with Dendrite Morphology for Classification

Recent neurological studies have shown the importance of dendrites in neural computation. In this paper, a neuron model with dendrite morphology, called logic dendritic (LDNM), is proposed for classification. This consists four layers: synaptic layer, membrane and soma body. After training, LDNM simplified by proprietary pruning mechanisms further transformed into circuit classifier. Moreover, to address high-dimensional challenge, feature selection employed as dimension reduction method before training LDNM. addition, effort employing heuristic optimization algorithm learning also undertaken speed up convergence. Finally, performance assessed five benchmark classification problems. comparison other six classical classifiers, achieves best on two (out five) The experimental results demonstrate effectiveness model. A new perspective solving problems provided paper.