ML tek5

Repository of the Tek5 machine learning course at Epitech, year 2025-2026.

The first session starts September 18th 2025

The second session starts November 27th 2025

Python environment

To manage your python environment, several solutions exist.

uv is more modern and should work faster than pip.

A requirements.txt is at the root of the repo.

uv

This page contains instructions on managing a project with uv and installing from requirements.txt.

https://docs.astral.sh/uv/guides/projects/

pip

Classic python package manager.

https://docs.python.org/dev/installing/index.html

Repo organisation

code/

Examples, and support for exercises done during the course.

documents/

Useful documents and references.

slides/

All the pdf presentations of the course.

project/

Description of the projet and required datasets.

Content of the course

The course is given over 5 days. Please note that during the first session, the organisation might evolve a little bit.

Day 1

• Introduction

• Notion of train and test set, first regression examples

• Presentation of the project

Day 2

• Classification and logistic regression

• Technical prerequisites: probabilities, statistics, metrics

• Clustering

• Validation of the datasets chosen by the students for the project

Day 3

• Optimization and gradient-based algorithms

• Neural networks

• Scoring

• Project follow-up

Day 4

• Dimensionality reduction

• Density estimation

• Feature selection

• Project follow-up

Day 5

• Classification and regression trees, ensemble methods

• Reinforcement learning

• Simplicity bias of neural networks.

• Project follow-up