rencccc / printed-digits-classification Goto Github PK
View Code? Open in Web Editor NEWThis project forked from fruitzzx/printed-digits-classification
This project presents a hybrid algorithm for training RBF network based on K-means and SOM. The algorithm consists of a proposed clustering algorithm to position the RBF center and givens least squares to estimate the weights. The aim of this experiment is to recognize printed digits (1-4) using the hybrid model. In the meanwhile, KNN and MLP with Scaled Conjugate Gradient will be implemented in order to show the comparative of different models according to the experiments.