Implemented Recursive Neural Networks(RNN) and trained it using forward and back propagation algorithms to make a handwritten digit recognizer package in MATLAB with an overall accuracy of 94.3%
These files implement Handwritten digit recognition using Recurseive Neural Networks for a total of 3 layers with 1 hidden layer. The files can be run through the ex4 script. Backpropgation and Forwardpropgation were used to train the network.
Overall accuracy is about 95%
Implemented Recursive Neural Networks(RNN) and trained it using forward and back propagation algorithms to make a handwritten digit recognizer package in MATLAB with an overall accuracy of 94.3%