Hi!

I’m a PhD student at KTH Royal Institute of Technology, Stockholm. I use my background in information theory and error-correcting codes to analyse deep learning systems. I’m passionate about developing more efficient learning algorithms, as well as understanding why they work. This helps us democratise AI – not just by making tools more accessible, but by designing systems which we can trust and understand.

Recently, I’ve been exploring how theoretical tools can help us design the “right” assumptions (or inductive biases) so we can start to open up black-box algorithms and see what really drives their performance.

I’m fortunate to be supervised by Ragnar Thobaben and Mikael Skoglund. Before starting my PhD, I did both my MSc and BSc in electrical engineering at KTH. I also spent a year as an exchange student at Imperial College London, where I wrote my MSc thesis with Deniz Gündüz at the Information Processing and Communications Lab.

If any of this sounds interesting, I’d love to hear from you!

Papers

A Coding-Theoretic Analysis of Hyperspherical Prototypical Learning Geometry

Martin Lindström, Borja Rodríguez-Gálvez, Ragnar Thobaben, and Mikael Skoglund

GRaM Workshop @ ICML 2024 (PMLR 251)

arXiv BibTeX

Teaching Experience

Miscellaneous