Repository containing functions for binary classification, trained to predict Higgs Boson precence from particle decay signatures. Ioannis Bantzis (ioannis.bantzis@epfl.ch), Manos Chatzakis (emmanouil.chatzakis@epfl.ch), Maxence Hofer (maxence.hofer@epfl.ch)
This repository consists of the following files:
- implementations.py: Contains all the implementations of the regression models implemented for this work.
- helpers.py: Contains some helper functions provided by the course
- utils.py: Utility functions, implemented to aid the implementation of the models
- run.py: Script reproducing the .csv file of our Aicrowd results. This is also the implementation of our partial training method.
- Usage:
python3 run.py
- evaluation.py: Script for evaluating the regressions models through cross validation.
- Example Usage:
python3 evaluation.py gd #cross validation for mean square gradient descent- To see detailed usage examples:
python3 evaluation.py usage #prints a helping prompt - report.pdf: A two-page report about this work.
- .gitignore: A basic gitignore file for python projects. It ignores /dataset/ named directories, in order to be able to store the dataset locally at our root directory.
- README.md: This readme.
We use the CERN particle signature dataset, available at https://www.aicrowd.com/challenges/epfl-machine-learning-higgs
This work was developed for the first project of the postgraduate course "cs433-Machine Learning" course of EPFL, during the fall semester of 2022.