Welcome to my website!

Hi, my name is Katja Hauser.

I finished my master’s thesis in Applied Computer Science at Visual Learning Lab (Heidelberg University) on Invertible Neural Networks in December 2019.

Currently, I am working as independent researcher. I am interested in understanding the structures that arise in neural networks during training. Specifically, I want to study the networks uncovered by the Lottery Ticket Hypothesis with topological methods.

I have previously worked with:

• invertible normalizing flows, specifically the Invertible Neural Network (INN)
• ensembles and ensemble based uncertainty quantification (specifically Deep Ensemble)

My CV can be found here.

You can find me on github. I also have a Twitter account, if you like to follow me there for updates on my work. I otherwise hardly use it, though.

If not indicated otherwise, feel free to share and use the content on my website under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. If you are interested in commercial use, contact me.