Connectionists: 4 postdoc positions, 1 PhD position and 1 scientific coordinator position to study Abstract Representations in Neural Architectures in humans and machines

[Gemma Roig]

Hello community!

We are happy to announce that the DFG funded research unit ARENA - Abstract Representations in Neural Architectures, situated at Goethe University Frankfurt, the Frankfurt Institute of Advanced Studies (FIAS) and the Max Planck Institute for Software Systems Saarbrücken, invites applications for
 
4 Postdoc positions (f/m/d) (E 13, full time),
1 PhD positions (f/m/d) (E 13, full time) and
1 position as scientific coordinator (f/m/d) (E 13, 50 % part time)
 
Preferred starting dates are between February and April, 2023. The duration of contracts will be up to four years. The salary grades will be based on the rules of the institution, at which the respective position is located.
 
ARENA is an interdisciplinary research unit funded by the German Research Foundation (DFG), with seven research projects from the fields of Experimental Psychology, Cognitive Neuroscience, Artificial Intelligence, and Neural Data Science. Our joint research goal is to combine experimental and modeling expertise to improve understanding of how abstract knowledge representations emerge and are coded in different neural architectures - in human brains and in biologically inspired AI models. To this end, we will explore how AI models can be used to better analyze experimental human data - behavioral experiments as well as eye tracking and brain imaging data - and in turn how such human data can be used to improve AI models. Find out more about ARENA at http://neuroai-arena.github.io/ and apply to join our interdisciplinary team to solve these exciting research questions!
 
We are currently seeking enthusiastic and highly motivated team members for the following projects: 
 
(1)  Bridging Levels of Abstraction in Brains and Natural Language Processing Machines. Project led by Dr. Mariya Toneva, Max Planck Institute for Software Systems, Saarbrücken (PostDoc position; E13-TV-L, full time, 2+2 years)
Although recent deep NLP models have seen a lot of success, they still exhibit several limitations. We posit that these are in part due to the formation of suboptimal abstract representations of language, which are important for faster learning and improved generalization to new tasks. This project will characterize and bridge the levels of linguistic abstraction between human brains and deep NLP models by combining large-scale neuroimaging datasets, behavioral data, and computational modeling. The ideal candidate should have experience analyzing brain recordings (fMRI, MEG) obtained from language stimuli, strong computational and programming skills (python, pytorch/tensorflow), and some experience with training transformer models. For further information please contact Mariya Toneva at mtoneva at mpi-sws.org <mailto:mtoneva at mpi-sws.org>. 
 
(2)  Emergence of Abstract Representations in Contextualized Multimodal Models. Project led by Prof. Dr. Gemma Roig, Department of Computer Science, Goethe University Frankfurt (PhD position; E13-TV-G-U, full time, 3+1 years)
The goals of this project include developing multimodal deep neural networks by integrating different perceptual modalities; to analyze and study the formation of abstract representations in those models, and then, use those models as computational models to understand how abstract representations are formed in the brain. Close interactions with all other projects within the ARENA unit are expected. The ideal candidate should have strong knowledge in machine learning (deep learning, RNN, LSTMs, Transformers, multimodal models), excellent programming skills in python, and extensive experience in Tensorflow, pyTorch or related libraries; knowledge of high performance distributed computing with GPUs. For further information please contact Gemma Roig at roig at cs.uni-frankfurt.de <mailto:roig at cs.uni-frankfurt.de>
 
(3)  Hierarchical Organization of Distributed Semantic Knowledge in the Human Language System. Project led by Prof. Dr. Christian Fiebach, Department of Psychology, Goethe University Frankfurt (PostDoc position; E13-TV-G-U, full time, 4 years)
The nature and format of semantic representations in the human brain have so far not yet been fully understood. This project aims at investigating in depth the underlying computational principles as well as neuroanatomical localization of lexical-semantic and conceptual-semantic representations at various levels of abstraction, by combining functional MRI and magnetoencephalography with embedding and other NLP models. A main goal of this project is to establish a comprehensive large-scale human neuroimaging dataset, which will be shared within ARENA and publicly. The ideal candidate has expertise in human functional neuroimaging, strong statistical and computational skills (data analysis with python, matlab, or similar tools), and should be motivated to learn and apply computational modeling. For further information please contact Christian Fiebach at fiebach at psych.uni-frankfurt.de <mailto:fiebach at psych.uni-frankfurt.de>. 

(4)  The Hierarchical Structure of Abstract Scene Representations; project led by Prof. Dr. Melissa Võ, Department of Psychology, Goethe University Frankfurt (PostDoc position; E13-TV-G-U, full time, 4 years)
The aim of this project is to understand the emergence and composition of abstract representations of objects in real-world scenes which allow for efficient behavior. We intend to pursue this aim by systematically probing the existence of a hierarchical structure in scene grammar using different input modalities and different tasks. Thanks to the multi-disciplinary composition of ARENA, we will be able to run a diverse set of studies, joining forces with our colleagues from computer science. While the PostDoc candidate should bring along strong skills with regard to state-of-the-art EEG and/or eye tracking in VR methodologies, great interest in acquiring new computational skills will also be key. For further information please contact Melissa Võ at mlvo at psych.uni-frankfurt.de <mailto:mlvo at psych.uni-frankfurt.de>.  
 
(5)  Navigating Knowledge Spaces Using Cognitive Maps; project led by Prof. Dr. Matthias Kaschube, Frankfurt Institute for Advanced Studies (PostDoc or PhD position; E13-TV-H, full time, 4 years)
The goal of this project is to explore the computational benefits of low- dimensional cognitive maps in brains and artificial neural networks. For this purpose, the proposal links the concept of cognitive maps with state-of-the-art generative models, such as, for instance, GANs. Within ARENA, we will study this in the context of category learning as well as learning the statistics of object co-occurrences in visual scenes using computational models that are tightly linked to behavioral experiments in humans. The ideal candidate should have strong computational and programming skills (python, pytorch/tensorflow), some experience with deep generative models and with the analysis of brain imaging data. Close interactions with all other projects within the ARENA unit are expected. For further information please contact Matthias Kaschube at kaschube at fias.uni-frankfurt.de <mailto:kaschube at fias.uni-frankfurt.de>.
 
Candidates with interdisciplinary expertise at the intersection of cognitive science, neuroscience, and computer science – or a high motivation to extend their knowledge in this interdisciplinary field – are particularly welcome. PostDoctoral candidates should have a University degree and a very good doctorate in the specified or related fields, as well as expertise in experimental techniques and/or advanced data analysis methods. For doctoral positions, applications are invited from students with an excellent master’s degree in computer science, AI, or related disciplines. All applicants must be fluent in English.
 
We are furthermore seeking
 
(6)  A Scientific Coordinator (E 13 TV-G-U, part time 50%) who will support the research conducted in the unit by a number of centrally coordinated measures, including (a) internal, national, and international scientific cooperation and networking (organization of colloquia, retreats, guest visits), (b) training and career development of early career researchers (workshops, equal opportunity measures, lab visits, etc.), (c) research data management and quality assurance, (d) reporting, and (e) strategic development of the research unit. The ideal candidate for this position has a scientific background (University degree and PhD in computer science, psychology, neuroscience or a relate area), high skills in organization and communication, very good strategic and analytical abilities, as well as very good English skills. Experience with or interest in research data management and open science are a plus. The scientific coordinator will closely collaborate with the speaker of the unit, Prof. Christian Fiebach, and all principle investigators of the ARENA unit. For further information please contact Christian Fiebach at fiebach at psych.uni-frankfurt.de <mailto:fiebach at psych.uni-frankfurt.de>.
 
The ARENA research unit is a collaborative project of Goethe University Frankfurt, Germany, the Frankfurt Institute of Advanced Studies, both part of a vibrant cognitive science/neuroscience community, and the Max Planck Institute for Software Systems in Saarbrücken, Germany, one of the leading institutes in human-centered AI. ARENA researchers will have access to state-of-the-art research facilities, including laboratory facilities for human behavioral and brain imaging studies and dedicated GPU computing resources. Further details can be found at http://neuroai-arena.github.io/.
 
Goethe University, FIAS, and the Max Planck Institute are committed to a policy of providing equal employment opportunities for both men and women alike, and we therefore encourage particularly women to apply for the position/s offered. Individuals with severe disability will be prioritized in case of equal aptitude and ability.
 
If you are interested in joining ARENA, please send a CV, a statement of research interests, one (PhD positions) or two (PostDoc positions) publications, and the names and contact details of at least two academic references (all in one pdf document) until 11.01.2023 to neuroai.arena at gmail.com <mailto:neuroai.arena at gmail.com>. In your application, please specify the project(s) you are particularly interested in and cc your application to the respective principle investigators specified above. 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.srv.cs.cmu.edu/pipermail/connectionists/attachments/20221221/54983ec3/attachment.html>