Neural networking journal club
Neural networking provides a time and place for people to discuss and learn about current topics in machine learning, especially in - but not limited to - the fields of astronomy and cosmology. Each session will (hopefully) begin with a talk about current and upcoming work from users of machine learning. Time permitting, afterwards, there will be a discussion on other new techniques and interesting papers.

We are always looking for more speakers so if you have been using machine learning and want to present your work please email me at (even if the result has not been successful - it's always good to know what to avoid as well as what works).

To join the machine learning journal club mailing list, send an email with no subject and the word "subscribe" (without the quotation marks) in the main body of the email to

Normally, the sessions will take place on every other Wednesday at 14.30 in salle 281:

Upcoming neural networking journal clubs
Date Presenter Title
Beginner's guide to neural networks
When I arrived at the IAP I gave two introductory talks on how machine learning works (see picture above). The slides for these two session are attached here. The first session really addressed the basics and introduced some of the images which are likely to be seen in machine learning papers, whilst the second session had a collections of hints and tricks for improving machine learning. The second session also has a Jupyter notebook with examples from my work on classifying supernovae using machine learning.

Click here to download the slides from the 08/06/2017 neural networking session.

Click here to download the slides from the 06/07/2017 neural networking session. If you want to download the Jupyter notebook with the data so that you can run everything you can click here.
Interesting papers
Here is a short list of papers which make use of machine learning that I have found particularly interesting (this doesn't mean that if a paper is not on the list then it is not interesting - I might have just missed it). I will keep the list regularly updated whenever something piques my interest. If anyone spots anything they think I should see send me an email to
Optimizing spectroscopic follow-up strategies for supernova photometric classification with active learning
E. E. O. Ishida, R. Beck, S. González-Gaitán, R. S. de Souza, A. Krone-Martins, J. W. Barrett, N. Kennamer, R. Vilalta, J. M. Burgess, B. Quint, A. Z. Vitorelli, A. Mahabal, and E. Gangler
Return of the features - Efficient feature selection and interpretation for photometric redshifts
A. D’Isanto, S. Cavuoti, F. Gieseke, and K.L. Polsterer
Automatic physical inference with information maximising neural networks
Tom Charnock, Guilhem Lavaux, and Benjamin D. Wandelt
Generating and refining particle detector simulations using the Wasserstein distance in adversarial networks
Martin Erdmann, Lukas Geiger, Jonas Glombitza, and David Schmidt
Automatic survey-invariant classification of variable stars
Patricio Benavente, Pavlos Protopapas, and Karim Pichara
Uncertain classification of variable stars: Handling observational gaps and noise
Nicolás Castro, Pavlos Protopapas, and Karim Pichara
Fast Cosmic Web Simulations with Generative Adversarial Networks
A.C. Rodríguez, T. Kacprzak, A. Lucchi, A. Amara, R. Sgier, J. Fluri, T. Hofmann, and A. Réfrégier
Cosmic String Detection with Tree-Based Machine Learning
A. Vafaei Sadr, M. Farhang, S.M.S. Movahed, B. Bassett, and M. Kunz
Three Factors Influencing Minima in SGD
Stanisław Jastrzębski, Zachary Kenton, Devansh Arpit, Nicolas Ballas, Asja Fischer, Yoshua Bengio, and Amos Storkey
Astrophysical Data Analytics based on Neural Gas Models, using the Classification of Globular Clusters as Playground
Giuseppe Angora, Massimo Brescia, Giuseppe Riccio, Stefano Cavuoti Maurizio Paolillo, and Thomas H. Puzia
Muon Trigger for Mobile Phones
M Borisyak, M Usvyatsov, M Mulhearn, C Shimmin, and A Ustyuzhanin
An automatic taxonomy of galaxy morphology using unsupervised machine learning
Alex Hocking, James E. Geach, Yi Sun, and Neil Davey
Uncertainties in parameters estimated with neural networks: application to strong gravitational lensing
Laurence Perrault Levausseur, Yashar D. Hezaveh, and Risa H. Wechsler
Photometric redshift estimation via deep learning
Antonio D'Isanto, and Kai Lars Polsterer
Creating Virtual Universes Using Generative Adversarial Networks
Mustafa Mustafa, Deborah Bard, Wahid Bhimji, Rami Al-Rfou, and Zarija Lukić
Hyperparameter Optimization: A Spectral Approach
Elad Hazan, Adam Klivans, and Yang Yuan
I try and keep a more complete list here.
Previous neural networking sessions
Date Presenter Title
11th April 2018
Salle 281
Alessandro Manzotti Mixture density networks
29th March 2018
Salle des séminaires
Thomas Keck Machine learning workshop at the IAP 2018
28th March 2018
Salle 281
Thomas Keck Machine learning for the Belle-II experiment
7th March 2018
Salle des séminaires
François Lanusse Variational auto-encoders
7th February 2018
Salle 281
Florian Führer Doing maths with machine learning
6th December 2017
Salle 281
Open discussion Bring your questions along, have a mince pie, feel festive!
22nd November 2017
Salle 281
Tom Charnock Taking the square
(Information maximising non-linear data compression)
8th November 2017
Salle 281
Marc Huertas-Company Deep-learning for galaxy evolution
26th October 2017
Salle 281
Jean-Marc Delouis Unsupervised classification based on scattering transform and t-SNE
11th October 2017
Salle 281
Farhang Habibi Object classification and analysis for SDSS DR12
15th September 2017
Salle des séminaires
Nino Vieillard Measurement of nearby galaxies' morphology by deep learning
6th July 2017
Salle des séminaires
Tom Charnock Tips, tricks and some classification using neural networks
8th June 2017
Salle des séminaires
Tom Charnock Beginner's guide to neural networks