Machine-Learning-Training

hpc-ai-ml-2022

This repository holds the computer labs for the Introduction to Machine Learning course of the 2021-2022 HPC-AI MSc https://www.hpc-ai.mines-paristech.fr/

Requirements

The labs were developed for Python3. All required packages are part of the Anaconda platform so you can simply install Anaconda3 on your machine. If you'd rather install just the required packages with pip, that is also possible. The labs were developed for Python 3.4.3, with the following libraries:

• numpy 1.16.5
• scipy 1.2.2
• matplotlib 2.2.4
• pandas 0.22.0
• seaborn 0.9.0
• sklearn 0.19.2

Quick start

Clone the 2022-mines-HPC-AI-TD repository :

git clone https://github.com/NelleV/2022-mines-HPC-AI-TD.git

Create the conda environment tp-ml :

cd 2022-mines-HPC-AI-TD
conda env create -f environment.yaml

Activate the environment tp-ml :

conda activate tp-ml

To show the conda environment in Jupyter Notebook, register the kernel:

python -m ipykernel install --user --name tp-ml --display-name "TP ML"

ps : to unregister the kernel:

jupyter kernelspec uninstall tp-ml

To check your installation, try launching Jupyter (e.g. by typing jupyter noteboook in a shell), navigating to where you have downloaded/pulled the first lab (.ipynb file) and opening it. In that notebook (or in a python terminal), you should be able to run

  import sklearn
  import pandas
  import seaborn
  import matplotlib

Program

• Lab 1: Principal Component Analysis (1h)
• Lab 2: Data normalization (1h)
• Lab 3: Cross validation & features
• Lab 4: Linear and logistic regression (1h)
• Lab 5: Regularized models (1h)
• Lab 6: Trees and Forests
• Lab 7: k-NN

Acknowledgements

These notebooks are adapted from notebooks previously created with the help of Chloé-Agathe Azencott, Judith Abecassis, Joseph Boyd, Arthur Imber, Benoit Playe and Mihir.