Machine Learning Workshop

Workshop Materials

You can download all datasets and Jupyter Notebooks used in this workshop by following this link.

To run the tutorials below, you need to have Jupyter Notebook installed. You can get it easily using anaconda.

Below, we list the notebooks for each tutorial and the required packages. Note that some of these are already installed by default and you do not need to install them again. For installing the remaining packages, you can use "pip install" (download pip here) or "conda install" (if you have anaconda installed).

Tutorial 1

Notebooks

  • tutorial_intro.ipynb (introduction to CNN with regression as example.)
  • tutorial1a.ipynb (Example of galaxy morphology classification using CNN on the CANDELS/HST H-band.)
  • tutorial1b_TSNE.ipynb (Example of galaxy morphology classification using T-SNE. This tutorial also shows how to explore the dataset in terms of other physical properties of the galaxies.)
  • tutorial1c_SOM.ipynb (Example of galaxy morphology classification using self-organizing maps (SOM). This tutorial also shows how to explore the dataset in terms of other physical properties of the galaxies.)
  • tutorial_1d.ipynb (Visualization of CNN filters and layers.)

Requirements/Packages

  • os, time, sys, random
  • matplotlib
  • numpy
  • pandas
  • pickle
  • tensorflow
  • keras
  • sklearn
  • astropy
  • cv2
  • scipy
  • mpl_toolkits
  • regions
  • pymvpa2

Other Materials

  • Embedding Projector for visualization of datasets (e.g., using T-SNE, UMAP, PCA, etc). See also tutorial_1d.ipynb.

Tutorial 2

Notebooks

  • tutorial2_IRAC.ipynb (Analysis and reduction of Spitzer Space Telescope data using Random Forest.)

Requirements/Packages

  • time
  • matplotlib
  • numpy
  • pandas
  • statistics
  • xgboost
  • scipy
  • sklearn
  • eli5
  • seaborn
  • pdpbox

Other Materials

  • ...

Tutorial 3

Notebooks

  • tutorial3a_LinearRegression.ipynb (Introduction to Bayesian Neural Networks on the example of linear regression.)
  • tutorial3c_BNNforGWparameters.ipynb (Extraction of physical parameters a black hole / neutron star merging event from LIGO data using Bayesian NNs.)
  • tutorial3b_DeNoiseLigoData.ipynb (Denoising of LIGO gravitational wave data using Bayesian NNs.)

Requirements/Packages

  • os, sys, random, string
  • matplotlib
  • h5py
  • pathlib
  • pickle
  • numpy
  • tensorflow
  • tensorflow_probability
  • seaborn
  • torch
  • scipy

Other Materials

  • ...