Handwritten digits recognition using ensemble neural networks and ensemble decision tree

Handwriting recognition is widely used, and the using of neural network as a method to do is quite common. In this project, neural networks ensembles combined with another classifier are train and test in solving handwritten digit recognition problems, using USPS and MNIST database. The new proposed algorithm, ensemble neural networks that combined with ensemble decision tree (ENNEDT), performed better than single neural network and ensemble neural network. ENNEDT reached 84% accuracy from classifying USPS dataset. Matlab program implemented the training and testing functions to the handwritten digit recognising system.