Neural_Network

Neural Network implementation from scratch for classification task

• Place the data.pkl containing train, validation and test set in the datapath directory

• Modify the main appropriately based on the desired: hidden dimensions, activation function, normalization, number of epochs,...

• Run:

$ python NN.py

This will do the training and prediction, the result is a dictionary containing the accuracy and loss scores. For some examples of performance on SVHN dataset refer to the notebook: NN_plots.ipynb