Skip to content

This repository is based on the material produced during the "Machine Learning Course" at University of Bari for PhD first year students.

Notifications You must be signed in to change notification settings

bdanzi/ML_COURSEBARI

Repository files navigation

Contributors Forks Stargazers MIT LinkedIn

Table of Contents
  1. ML_COURSEBARI

ML_COURSEBARI

The NormDat_test.txt and NormDat_train.txt files contain 30.000 randomized and normalized examples, each having the first column with the true label and 23 particle physics features. This is used by a ML class in Bari for testing different Machine Learning algorithms and their performance

Darwin MacBook-Pro.local 21.4.0 Darwin Kernel Version 21.4.0: Fri Mar 18 00:47:26 PDT 2022; root:xnu-8020.101.4~15/RELEASE_ARM64_T8101 arm64

Authors

Requirements

$ pip install numpy
$ pip install sklearn
$ pip install matplotlib
$ pip install pandas

Instructions

In the ML_COURSEBARI\ directory of this repository:

$ python ml_exercise.py

The command will produce .png files:

  • ROC curves ROC_results.png
  • Loss function vs Trainsize numbertrainingsamples_loss_mse.png
  • MLP score vs Trainsize numbertrainingsamples_MLP_score.png

Examples of the plot produced are stored in the Plots_NNtraining\ directory for the usage of a Multi Layer Perceptron (MLP) and a Boosted Decision Tree (BDT) implemented by using scikit-learn package.

Your own Neural Network from scratch

$ python NN_backpropagation_bdanzi.py

This python code has been written to implement the backpropagation process in Neural Network training not using already available open-source packages. The command will produce the .png file:

  • Loss function vs Epochs This code represents a home-made (no usage of numpy, scikit-learn,keras) Neural Network having one hidden layer.

About

This repository is based on the material produced during the "Machine Learning Course" at University of Bari for PhD first year students.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published