Skip to content
Switch branches/tags

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time


Far-InfraRed Emission Networks (FIRE-net) is a machine learning framework that aims to estimate the far-infrared (FIR) spectral energy distribution (SED) of a galaxy, based on the ultraviolet to mid-infrared (UV-MIR) SED.

>>> Interactive plots can be found here <<<

This github repo provides the following:

  • jupyter notebooks that guide the process from raw data to a fully trained model
  • a jupyter notebook that shows how to apply our fiducial model quickly
  • a library of helper classes/functions
  • the DustPedia + H-ATLAS SED fitted data (about 23 MB)


If you want an example of how the neural networks are trained or used, see the jupyter notebooks in the notebooks directory. We recommend viewing them using nbviewer, although they can be opened directly from github as well.

The notebooks can also be run dynamically using binder. This avoids the need to set up an environment locally, since the code is run in the cloud.


Using the code locally

The environment that was used to run the notebooks can be built from the either environment.yml (conda) or environment.txt (pip). We strongly recommend using a "virtual environment": a separate python installation which does not interfere with your base environment. Possible options are conda, pipenv or virtualenv/venv.

Jupyter lab is the recommended tool to run the jupyter notebooks. You can install it in a separate environment (e.g. the base environment), and add your environment as separate kernel. In that case, you can remove jupyterlab from environment.yml. Alternatively, jupyter lab can be installed in the current environment and run from there.

For conda users:

conda env create -f environment.yml
conda activate firenet
jupyter lab

For pip users:

python3 -m venv firenet-env
source firenet-env/bin/activate
pip install -r requirements.txt
jupyter lab

Alternatively, manually install the missing packages from environment.yml into your favourite machine learning environment.


This work is accompanied by the paper "Predicting the global far-infrared SED of galaxies via machine learning techniques". The paper can be found here (arXiv pdf, full paper). If you use this work, please cite the paper. Following bibtex can be used:

       author = {{Dobbels}, W. and {Baes}, M. and {Viaene}, S. and {Bianchi}, S. and
         {Davies}, J.~I. and {Casasola}, V. and {Clark}, C.~J.~R. and
         {Fritz}, J. and {Galametz}, M. and {Galliano}, F. and {Mosenkov}, A. and
         {Nersesian}, A. and {Tr{\v{c}}ka}, A.},
        title = "{Predicting the global far-infrared SED of galaxies via machine learning techniques}",
      journal = {\aap},
     keywords = {galaxies: photometry, galaxies: ISM, infrared: galaxies, Astrophysics - Astrophysics of Galaxies},
         year = 2020,
        month = feb,
       volume = {634},
          eid = {A57},
        pages = {A57},
          doi = {10.1051/0004-6361/201936695},
       eprint = {1910.06330},
       adsurl = {},


Far-InfraRed Emission Networks: Using ML techniques to estimate a galaxy's FIR SED.




No releases published


No packages published