#Simple Backward Propagation Neural Network
A simple implementation of the neural network with one hidden layer using c++.
The program could modify:
- number of input/hidden units
- learning rate
- the initial weight setting
And the settings are:
- both hidden/output units use the sigmoid function
- the standard squared error is used
- the input, network output, target output, and error are shown in each iteration
The program has a Makefile. Simply type make
, and it will compile for you!
To run the program: execute with ./backprop
To run with your own datafile: put your datafile in the same folder of the source codes and change its name to trainingData.txt, and re-run the above.