Project 1 for PCML 2016. Kirtan Padh, Louis Baligand, Pierre Runavot
This is a repository for the course Pattern Classification and Machine learning offered at EPFL in Autumn 2016. We use the dataset provided in https://www.kaggle.com/c/higgs-boson to classify the given decay signatures into 'Higgs Boson' and 'Not Higgs Boson'. We were not supposed to use any external machine learning libraries and use only functions implemented by ourselves.
costs.py, helpers.py & proj1_helpers.py: useful general methods(compute loss, load data, batch iter...). implementations.py: the six basic methods seen in class run.py: scripts which produces the .csv of our optimal estimation