• Python > 3.5
• Keras >= 2.1.0
• sklearn >= 0.19
• jupyter
• sqlite3

(the code may run with earlier version)

All the data ued in this work are provided on the data directory.

To get the same results as the one presented on the report you just have to run the python scripts in the src directory.

To run the perceptron with dimension reduction:

python3 -m src.perceptron

To run the perceptron with dimention reduction and clustering features:

python3 -m src.clustered_perceptron

Both of them will generate .csv files in the stats directory. This to files need to be loaded into the plottre/data.db sqlite database in order to be used correctly by the different notebook.

You will find more information on the database schema on first cell of the different notebooks.

The four part of this work are divided on four different notebooks:

• Clustering: plotter/dataset.ipynb
• Dimension reduction plotter/dimesion_reduction.ipynb
• Perceptron with dimension reduction plotter/nn.ipynb
• Perceptron with dimension reduction and clustering plotter/clusper.ipynb

As said above, the last two need the results to be computed and loaded into the database.