We have four open positions for PhD students or postdocs in the Neural Data Science Lab at the Campus Institute Data Science (University of Göttingen) and the Max Planck Institute for Dynamics and Self-Organization in Göttingen, Germany.
If you want to work with us on exciting projects developing machine learning methods for neuroscience, cognitive science and physics, please check out the three short project descriptions below and apply now. Details can also be found on our website:
https://eckerlab.org/applications/
Application documents including a short motivation letter, CV, transcript of records (PhD students) and the contact details of two references should be sent to Alexander Ecker:
ecker@cs.uni-goettingen.de
We value diversity and equality. We therefore particularly encourage international and female applicants as well as applications from underrepresented groups or candidates with disabilities.
============
*(1) Data-driven multi-modal discovery of cell types in the neocortex*
Understanding the relationship between structure and function of cortical neurons and circuits is one of the key challenges in neuroscience. In this project, we develop deep learning methods for data-driven identification of excitatory cell types in the visual
cortex and to understand how a neuron’s morphology relates to its function. We will harness a unique large-scale functional anatomy dataset: a combination of electron-microscopy reconstructions at sub-micrometer resolution with two-photon functional imaging
of nearly all excitatory neurons in one cubic millimeter of the mouse visual cortex.
Role: Postdoc (preferred) or PhD student
Your profile: Background in systems / computational neuroscience or machine learning (deep learning in particular). Experience with both is a strong plus, but not a requirement.
This project is funded by the European Research Council through an ERC Starting Grant and in tight collaboration with Andreas Tolias' lab at Baylor College of Medicine, Houston, TX, USA. Regular visits at our collaborator’s lab in Houston are possible/encouraged.
============
*(2) Action Capture Platform: Deep learning for movement analysis and action classification*
Studying social interactions and the underlying cognitive processes in humans and nonhuman primates requires observing and quantifying their behaviour, including body and head orientation, both in interactive lab settings and in the wild. The long-term goal
of this project is to develop methods for automated, video-based recognition, motion tracking, and pose estimation of individual nonhuman primates as well as recognition of action types and inter-agent interactions in groups of primates freely moving primaates.
We strive for generalisation across a range of primate species and varying real-world environmental conditions without the need to acquire detailed keypoint annotations in every context. The automated behavioural analysis tools developed in this project constitute
a key enabling technology for a Collaborative Research Center focused on studying the Cognition of Interaction.
Role: Postdoc (preferred) or PhD student
Your profile: Strong background in machine learning (deep learning in particular) and/or computer vision. Experience with congitive science is a strong plus, but not a requirement.
This project is funded by the German Research Foundation (DFG) as part of Collaborative Research Center SFB 1528 – Cognition of Interaction and in collaboration with several groups at the German Primate Center: Julia Ostner, Claudia Fichtel, Julia Fischer
============
*(3) Holography reconstruction using deep learning*
Digital in-line holography is a very powerful method to investigate particles in turbulent fluid flows such as atmospheric clouds, aerosols, etc. using the diffraction patterns of the particles illuminated by coherent light. Currently, wider use of the technique
is held back both by a data processing bottleneck and the degradation in hologram quality in field measurements compared to laboratory experiments due to the physical environment. In this project, we will train machine learning algorithms to reconstruct holograms
using a combination of real holograms from the lab and the field and simulated datasets. We will start with simple idealized Lorenz-Mie theory, but later extend to full wave-equation calculations including non-flat and non-Gaussian beams, realistic defects
in the optical components, dirty optics, camera micro-lens arrays, pixel cross-talk and digitization errors, non-spherical particles including ice crystals, aggregate particles with different indices of refraction, non-perfect collimation, non-perfect alignment,
spatial variation in the index of refraction of the air in the view volume due to temperature gradients, multi-scattering from multiple particles, etc.
Role: Postdoc (preferred) or PhD student
Your profile: Strong background in physics. Experience with machine learning (deep learning in particular) and computer vision is a strong plus, but not a requirement.
This project is in collaboration with the lab of Eberhard Bodenschatz and funded by the Max Planck Institute for Dynamics and Self-Organization.
============
--
Alexander Ecker
Professor of Data Science
Department of Computer Science • Campus Institute Data Science
University of Göttingen • Max Planck Inst. for Dynamics and Self-Organization
Goldschmidtstr. 1 • room 2.137 • 37077 Göttingen
https://eckerlab.org • @alxecker • +49 551 39-21272