Multilayer Perceptron

This is a Codam project. See subject.pdf for more details

Evaluation

You can go to mlp.ipynb to find the main part of the project. There you can load the dataset, prepare the data, choose model configuration, and train the model. Later you can see the evaluation of the model's performance.

There is no need to rerun the notebook but you can. You need python to run the Jupyter notebook. If you need help with that, go here: https://python.land/data-science/jupyter-notebook

Also you should check classes directory for the main "under the hood stuff".

• layers.py has Active class that handles feed-forward and backpropagation parts of the neural network, as well as weight initialization and stuff like that;

• model.py is about keeping track of macro parameters and orchestrating the learning and predicting phases. Also, displaying graphs.