Run this tutorial using Google Colab
This tutorial demonstrates some simple usecases of machine learning and deep learning for astrophysicians. It was first showcased during the SFtools-Bigdata workshop in november 2020.
Part 1 shows how to use scikit-learn to train shallow statistical models such as Support Vector Machines (SVM) and Random Forests on tabular data for star type classification based on their physical properties (temperature/radius/luminosity).
Part 2 demonstrates how to work with unstructured data such as images. It moves gradually from hand-crafted features (histogram of gradients) to learnt features using deep convolutional networks.
Part 3 gives examples of nice party tricks achievable using deep features such as clustering in 2D space, image retrieval, fine-tuning pretrained networks and so on.
Feel free to open an issue or a pull request if you find any error or problem in this code.