Neural-Network-Sigmoid

Flexible Neural Network that can change the number of layers and nodes per layer. Only utilizes sigmoid activation. Used to recognize the handwritten digit set from kaggle.

Instructions:

1. Run "nn.py" file with Python
2. Specify the parameters when prompted (Training sample size, number of layers, nodes per layer, regularization parameters, etc.)
3. The backpropogation will run and print out the cost function every training step
4. The trained network will provide the test accuracy and run the network on the test dataset, saving the result to a csv file.