Search

link to homepage

Central Institute of Engineering, Electronics and Analytics (ZEA)

Navigation and service


Advertising division: ZEA-2 - Electronic Systems
Reference number: 2020M-032, Electrical engineering, physics

Master Thesis: Underlying Geometry in Reservoir Computing

The Central Institute of Engineering, Electronics and Analytics – Electronic Systems (ZEA-2) is a scientific engineering institute of Forschungszentrum Jülich GmbH performing research and development projects in close cooperation with the institutes of the research center as well as external partners. The focus of our work is electronic and information technology system solutions in sensor and detector technology, signal and data processing as well as measurement techniques. Hereby the preferred approach are highly integrated silicon-based System-on-Chip (SoC) solutions.

Project Description:
This project is running within the “Advanced Computing Architecture” project at Forchungszentrum Jülich GmbH, aiming towards multi-scale natural-density Neuromorphic Computing (www.fz-juelich.de/aca). Artificial recurrent neural networks (RNN) dominate a large class of computational models, having applications ranging from machine learning to modeling biological brain and computational neuroscience. Despite their widely acknowledged potential and computational power, RNN’s, however, remain insufficient during learning process. For overcoming this difficulties, reservoir computing has been proposed as a paradigm shift in RNN training. Unlike the full-system training of RNNs, the echo state computational substrate, i.e. the reservoir, would be randomly initialized and the output mask be the subject of training procedure. The mere idea of generating a reservoir at random, however, seems unsatisfactory; task-oriented dedicated reservoirs should lead to better results.

Your Tasks:

In the framework of this project, a theoretical scheme (alongside network simulations) shall be developed to enable the comparison between different task-oriented reservoirs, making use of the dynamical and topological properties (e.g. spectral dimension) of these reservoir systems.

Your Profile:

  • Studies in Mathematics or Physics as major part
  • Preferred knowledge in dynamical systems\network theory
  • Knowledge of programming (e.g. MATLAB or Python)
  • First experience in modeling neural networks (Mathematically and programming)
  • Willingness to teamwork is expected

Our Offer:

  • Opportunity to work together in a team of highly motivated scientists and technicians
  • Further development of your personal abilities in connection with a socially balanced work environment
  • Strongly supported on-the-job training at the start

Forschungszentrum Jülich GmbH aims to employ more women in this area and therefore particularly welcomes applications from women.

We also welcome applications from disabled persons.

Application
Did we raise your attention and you got curious?
We are looking forward for your application.
For further information please contact:

Forschungszentrum Jülich GmbH
Central Institute of Engineering, Electronics and Analytics
ZEA-2 – Electronic Systems
Britta Hallmann
52425 Jülich

E-Mail: verwaltung.zea2@fz-juelich.de


Servicemeu

Homepage