K. Seeliger
publications | contact | misc
About
Research
I am a postdoctoral researcher at the Max Planck Institute for Human Cognitive and Brain Sciences in Leipzig and a member of the Vision and Computational Cognition Group led by Martin Hebart. I'm also affiliated with Neural Coding Lab at Donders Institute in the Netherlands, where I did my PhD.
My research area is neural coding, in particular the similarities between representation hierarchies learned by convolutional neural networks and by biological visual systems. At the moment we are exploring what can be learned from training convolutional neural network models of the visual system directly on large-scale human fMRI data. Instead of training for object recognition first, such end-to-end modeling is directly learning to predict brain activity, providing us with an in-silico model of the visual system and allowing us to analyze its weights and behaviour to any input. You could also say we create copies of parts of the information processing of a real brain inside neural networks, and use them to study the visual system at large scale.
As this research provides us with maps of sensory feature detectors in the brain, one side result is improved ability to reconstruct perception, and maybe someday imagery and other content from brain activity.
Data and code sharing is common in machine learning but not yet in neuroscience. The Donders Institute has recently established a policy that all data and code will be shared upon publication. We are trying to publish code alongside papers on GitHub, but please contact me if you need something that is not there.
Education / experience
I did a BSc in Applied Computer Science at Bauhaus UniversitƤt Weimar and a MSc in Computational Neuroscience at Bernstein Center for Computational Neuroscience (BCCN) Berlin. I have obtained my PhD degree in the research group Artificial Cognitive Systems by Marcel van Gerven at the Donders Institute for Brain, Cognition and Behaviour in the Netherlands.
During university I have done lab rotations at CiNet, Berlin-BCI, TU Berlin ML group (thesis), hayneslab, MPI-MIS, Vernetzte Medien (thesis), Cybermedia Center and others. I used to work as a student assistant (data mining / processing) at the WebIS group and at DLR Asteroids & Comets (German Space Agency), as an ML intern for Datameer (Bay Area office) and as a localisation tester for Rockstar Games Lincoln. You can find my full CV (academic style) here.
I have been tutoring in the undergraduate courses Artificial Neural Networks, Python for AI and Scientific Writing.
Publications
The projects to which I contributed can be found on my Google Scholar profile.
Contact
katj posteo de (PGP key, vCard)𝕏 @seelikat
⊙ Leipzig, Koffer in Berlin.
Please let me know if you would like to grab a coffee / Club Mate in the Berlin / Leipzig area.
misc
Nice tools and pointers: Kagi search, Mendeley, pycortex, Detexify, BBEdit, vim, todo.txt, Mattermost, Calculating Space, mynoise & musicForProgramming.