From PCA to Autoencoders

Part of the course "Algorithms in Machine Learning" given in 2019 at ISAE-Supaero engineering school (filière SDD).

Course material

This repo contains following course material:

• slides-anim.pdf: lecture presentation
• slides.pdf: lecture presentation, without animations (lighter)
• slides/: slides source
• bibliography/: PDFs of cited reference papers
• MLAutoencoders.ipynb: practice notebook
• MLAutoencoders-student.ipynb: practice notebook, student version with some missing code
• MLAutoencoders-outputs.ipynb: practice notebook, with outputs already executed

Option 1: Running notebooks on Colab

The easiest option to get the notebooks running is Google Colab. The only prerequisite is having a Google account. Colab provides you a Python environment, all necessary libraries, and a (free!) compute infrastructure with CPU, GPU or TPU.

1. Open colab.research.google.com.
2. Import notebook from GitHub.
3. Type "FlorentF9" in the search bar and select the repo "Supaero-mlautoencoders".
4. Open a notebook, for example MLAutoencoders-student.ipynb.
5. Click "Copy to Drive" to be able to modify and save a copy the notebook.

Option 2: Running notebooks locally

The second option is to run the notebooks on your local environment. The required libraries are listed in requirements.txt. After having installed the dependencies in your favorite Python environment, simply clone this repo and open the notebooks in jupyter-notebook or jupyter-lab.