3 PhD student or Postdoc positions, on the subject of Architectures and learning methods for hierarchical cognitive processing

The Netherlands Institute for Neuroscience (NIN) and the Centrum Wiskunde & Informatica (CWI) have a vacancy in the Machine Learning research group for a talented

3 PhD student or Postdoc positions, on the subject of Architectures and learning methods for hierarchical cognitive processing.

Job description

Our recent theoretical insights have allowed a preliminary understanding of how animals learn to represent and memorize the key features of sensory stimuli for the guidance of action based on reinforcement learning (Rombouts et al. PloS Comp Biol 2015; Holtmaat & Roelfsema, Nature Reviews Neuroscience, 2018). The present project focuses on the hierarchical learning of cognitive subroutines, which can be incorporated into larger routines, by trial and error. We aim to develop a biologically plausible learning rules that endow neural networks with the possibility to learn hierarchical task structures and learn and execute cognitive tasks.

Topics of interest include - the role of learnable/flexible working memory models in learning and executing hierarchically structures routines. The project is a collaboration between in the Vision & Cognition group (Prof.dr. Pieter Roelfsema) at the Netherlands Institute for Neuroscience and the Machine Learning Lab at the Centrum Wiskunde & Informatica in Amsterdam (Prof.dr. Sander Bohte).

Further details:
3 PhD student or Postdoc positions, on the subject of Architectures and learning methods for hierarchical cognitive processing