Matthijs Mars

I am Matthijs and am a fourth year PhD student in the College for Doctoral Training in Data Intensive Science. I’m based at the Mullard Space Science Lab working with Jason McEwen and Marta Betcke on learned imaging methods with applications in astronomy. One of my recent projects involved designing learned reconstruction techniques for a concept design of an optical interferometer called SPIDER and I’m currently extending this to radio interferometry.

Besides that I have been involved in a group project with Spotify, precomputing audio features of podcasts to enhance the Spotify podcast dataset for use in the TREC Podcast track, a data challenge for information retrieval from podcasts.

Additionally, I spend 6 months this year working at the STFC Hartree Centre on developing deep reinforcement learning methods to control the shape of plasma in nuclear fusion reactors. This work was supervised by Adriano Agnello,George Holt and Nicola Amorisco.

I’m also a teaching assistant on the Machine Learning with Big Data course at University College London.

Publications

Using conditional GANs for convergence map reconstruction with uncertainties

Jessica Whitney, Tobías Liaudat, Matt Price, Matthijs Mars, Jason D. McEwen "Learned radio interferometric imaging for varying visibility coverage." ArXiv e-print, 2024.

Learned radio interferometric imaging for varying visibility coverage

Matthijs Mars, Marta Betcke, Jason McEwen, "Learned radio interferometric imaging for varying visibility coverage." ArXiv e-print, 2024.

FreeGSNKE: A Python-based dynamic free-boundary toroidal plasma equilibrium solver

Nicola C. Amorisco, Adriano Agnello, George Holt, Matthijs Mars, James Buchanan, Stanislas Pamela. "FreeGSNKE: A Python-based dynamic free-boundary toroidal plasma equilibrium solver." PoP, 2024.

Learned interferometric imaging for the SPIDER instrument

Matthijs Mars, Marta Betcke, Jason McEwen, "Learned Interferometric Imaging for the SPIDER Instrument." RASTI 2023.

Fast emulation of anisotropies induced in the cosmic microwave background by cosmic strings

Matthew A. Price, Matthijs Mars, Matthew M. Docherty, Alessio Spurio Mancini, Augustin Marignier, Jason. D. McEwen, "Fast emulation of anisotropies induced in the cosmic microwave background by cosmic strings." ArXiv e-print, 2023.

Audio Features, Precomputed for Podcast Retrieval and Information Access Experiments

Abigail Alexander, Matthijs Mars, Josh Tingey, Haoyue Yu, Chris Backhouse, Sravana Reddy, Jussi Karlgren, "Audio Features, Precomputed for Podcast Retrieval and Information Access Experiments." In the proceedings of Experimental IR Meets Multilinguality, Multimodality, and Interaction, 2021.