My group is using machine learning and data analytics for large-scale neuroscience, working on identifing cell types, circuits and computations in the healthy and diseased visual system. We are part of the Institute of Ophthalmic Research at the University of Tübingen and affiliated with the Center for Integrative Neuroscience and the Bernstein Center for Computational Neuroscience in Tübingen. Our official lab website can be found here.
This week, we have finally released the results of the spikefinder challenge to infer action potentials from calcium recordings. We had more than 50 submissions, many of them in the last few days of the challenge. Obviously, deep learning frameworks were hugely popular, but also “classical” generative modeling approaches (“MLspike”) did remarkably well. Here are the top 10:
We declared Patrick Mineault (Google), Nikolay Chenkov (BCCN Berlin), Peter Rupprecht and Stephan Gerhard (FMI Basel) joint winners, as there submissions were extremely close to each other. A lively debate emerged on twitter about what it all means:
— Philipp Berens (@CellTypist) 13. Mai 2017
Our preliminary analysis can be found on github. A more complete paper with thorough discussion and analysis will follow.
On occasion of the Meeting of the German Neuroscience Society, I received the FEI Technology Award 2017 for my work on using machine learning to define cell types in the nervous system.
Our paper on the functional diversity of bipolar cells and its origin in the inner retina has just been published online by Nature. We show that spatially extended stimulation induces a decorrelation between bipolar cell feature channels and that this effect is mediated by GABAergic amacrine cells. Richard Masland wrote a nice feature piece “Systems neuroscience: Diversity in sight“. The data will be available shortly.
Representing all authors, Katrin and I received the paper of the year award for a paper in the basic sciences from the medical faculty of the University of Tübingen 2016 for our paper “The functional diversity of mouse retinal ganglion cells“, awarded by Thomas Gasser, the vice-dean of the medical faculty.
Postdoctoral Researcher (TV-L E13 100%)
Medical Data Science for Ophthalmology
to be filled as soon as possible, with funding for up to 4 years.
I have been considering this for a while, here is one new year’s resolutions for 2017: Moving beyond Frontiers. There’s lots of evidence that we should not take Frontiers seriously (anymore?) as a scientific publisher.
Calcium imaging is one of the most widely used techniques to acquire data from large populations of neurons. Unfortunately, inferring action potentials from the measured calcium traces is not straightforward and algorithmically far from solved.
To make progress in this direction, we opened the spikefinder challenge at http://spikefinder.codeneuro.org/!
We collected a large dataset of simultaneous two-photon imaging and electrophysiologically records and made it available separated into training and test sets. We ask you to design your favorite algorithm and submit your results on the website!
What is your benefit?
- You get to advance the community
- All algorithms that beat state of the art will be included in an overview paper and you will be coauthor
- The top three teams will win a 1200/500/300 EUR award sponsored by our partner ZEISS AG.
We look forward from receiving your submissions. The deadline is the 31st of March 2017.
You might also be interested in the neurofinder challenge at http://neurofinder.codeneuro.org/ – a challenge to find the best ROI detection algorithm.
Philipp, Jeremy, Matthias, Andreas, Josh, Lucas
The first paper from my lab “Connectivity map of bipolar cells and photoreceptors in the mouse retina” has just been published by eLife. Congratulations to Christian for his first paper as well!
Together with Timm Schubert from Thomas Euler’s lab we reconstructed the photoreceptors in a publicly available EM dataset of retinal tissue. We build on the work of Moritz Helmstaedter and colleagues, who made their entire dataset and code available – to me this is a great example of how open science can fuel discovery. Our code and data can be found here. The project was our first adventure in connectomics, which provided a great learning experience for everybody involved.
Our collaborative research center “Robust Vision – Inference Principles and Neural Mechanisms” got approved by the German Research Foundation on Friday! This is great news – together with a great team of >20 PIs, we will tackle the question why biological visual systems show so remarkable robustness building on our joint expertise in experimental and computational neuroscience, machine learning and computer vision.