Our goal is to accelerate scientific discovery using machine learning and artificial intelligence: We want to develop computational methods that help scientists interpret empirical data and use them to discover and constrain theoretical models. To this end, we collaborate with experimental researchers from various disciplines. We are particularly interested in applications in the neurosciences: We want to understand how neuronal networks in the brain process sensory information and control intelligent behaviour, and to develop methods for the diagnosis and therapy of neuronal dysfunction.
More information about our research can be found on our publications and code pages.
Simulation-based inference Scientists build detailed models of the world, which is really really cool. But also, it generates a problem: it gets more and more difficult to tune the simulators such that they reproduce experimentally observed data. We use neural networks to fix this. Bam! Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. |
Blob detection We use deep learning to find the center of blobs. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. |
Low-dimensional dynamics We use VAEs and stuff to figure out if mice around running forward or backwards. Lots of cool stuff, missing data and so on. Also, uncertainty. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. |
Deep learning and the brain Small fruit flies are really dumb but they have great vision. We present bars to these flies and use a hexagonal CNN (boom!) to model this and super super cool we find that they are literally the same thing (mind blown boom boom!). Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. Hi This will be the text.Hi This will be the text.Hi This will be the text.Hi This will be the text. |
