| Software | Papers |
I did my PhD at the University of Bern with Walter Senn, Mihai A Petrovici and Katharina A Wilmes. Before that, I studied Systems and Computation (MSc) at the Institute of Neuroinformatics in Zürich. My interests lie in understanding computation in canonical neural circuits and recurrent neural networks.
With Alex Cayco Gajic, I study continual learning, context-dependent processing, and structured contextual representations in recurrent neural networks.
Independently, I develop compact parametric encodings of connectivity in neural networks, and propose them as efficient search spaces for neural architecture search through gradient-free optimization algorithms (e.g. evolutionary strategies). Central to these encodings are (i) the canonical nature of connectivity between (ia) neuron types and (ib) brain areas, and (ii) the assumption that connectivity is well captured by the interactions of a small number of genes.
Other interests include considering possibilities for societal change, cyberpunk and surreal animation cinema, writing fiction, and the japanese language.
Feel free to send me a mail if you want to discuss.