Neuron Classification Based on Semi-Supervised FCM Algorithm
Rong Ye Li, Xiao Jun Tong, Shan Zeng · Advanced materials research · 2013
The weighted Semi-supervised FCM Algorithm which is based on HPR multiplier (SSFCM-HPR) [1] is used for neuron classification in this paper. 231 groups of neuron data from NeuronMorpho.Org [2] are used in experiments as training samples and testing samples. The classification accuracy of testing samples can reach 93.90% in the experiment. The classification accuracies of testing samples are above 90% if the percentage of training samples is higher than 20%.