matlabRNN

An RNN implementation in Matlab

In this project you can train and test a fully functional RNN in Matlab. The network can have any amount of input neurons, output neurons, number of hidden layers (should be >= 2) and number of nodes per layer.

Usage:

1. Define default network parameters in mainRNN.m
2. Supply with training and test data (data must be normalized -1 to 1)
3. Choose a subset for cross evaluation data (optional)
4. Modify trainRNN.m to use your desired data instead

Or run as is to test on a rank 3 equation