This project provides a Python implementation of a Multilayer Perceptron. (MLP for short) neural network.
- Simple (but inefficient) implementation of a MLP.
- Application of the MLP to the well-known MONK dataset.
- File valentiMLP.py contains the implementation.
- File monk_simple.py contains the application to the MONK tasks.
The code is ready-to-go. MONK dataset is already provided in the "dataset" subfolder.
You'll need Python 3.X to run the scripts, as well as the following python libraries:
- numpy
- pandas
- matplotlib
- time
- pprint
- scikit-learn (sklearn)
- itertools
- random
- math
You can install them using your favourite packet manager for Python (such as pip).
Just run the monk_siple.py script into the Python interpreter, or use the ValentiMLP class inside your own project. You're ecouraged to explore different hyperparameters grids for the model selection, in order to do that you'll have to modify the param_grid variable at line 134 of monk_simple.py. You can also load different MONK tasks (or different dataset) by changing line 76 and 77 of monk_simple.py. If you decide to use a different dataset than MONK, you will probably have to change of the preprocessing part of the script.
NB: depending on your hardware, the model selection may take a while.
For any additional information about this project, you can email me at valentiandrea@rocketmail.com.
You're free (and encouraged) to make your own contributions to this project.
Just be nice, and respecful of each other!
All source code from this project is released under the MIT license.