This repository contains course materials for a two-day machine learning course at the MWGaia-DN doctoral school in Coimbra, Portugal (9th - 13th September 2024), created by Friedrich Anders and Emily Hunt.
Supervised learning techniques: Classification and regression, with a focus on neural networks and tree-based algorithms. Work examples include source classification, determining stellar parameters from spectra, and estimating photometric redshifts.
Unsupervised learning techniques: Clustering algorithms, dimensionality reduction methods (e.g. PCA, t-SNE, UMAP, self-organizing maps)
Interpretable ML & the future: Interpreting machine learning models using SHAP (SHapley Additive exPlanations). Including uncertainties in ML models. The future of ML/AI, including the potential of transformers.
The dependencies for each supervised ML notebook can be quite complicated, and depend on the notebook itself.
The dependencies needed for the unsupervised ML notebooks are in requirements.txt. It is strongly recommended that you use a fresh virtual environment (or open each one on a cloud service like Google Colab.)
After creating a virtual environment with the tool of your choice (e.g. with venv), you can install them locally with pip via:
pip install -r requirements.txt
Python 3.12 was used to develop this course. We recommend using the latest version of Python if possible - though (slightly) older versions should also be ok.
You may also be interested in some of the other course materials from this summer school, which can also be found here on GitHub: