Master Thesis - Online Unsupervised Learning in Brain-Inspired Neural Networks

Updated: about 1 year ago
Deadline: The position may have been removed or expired!

Your Job:

In the study presented in [3], the model was benchmarked using artificially generated sequences composed of tokens represented by a Latin character, which serves as a placeholder for any arbitrary objects. For example, a sequence could be composed of {‘A’,‘B’,‘D’,‘C’} for which the network needs to predict the next element in the sequence, say ‘D’, if presented successively with the elements ‘A’ and ‘B’. The way the model identifies the sequence elements is prewired, i.e., the presentation of a sequence element activates a specific subpopulation of neurons. This is problematic as the features or the elements of most real-world data are not known a priori, e.g., data streamed from an event-based camera [4]. In the present project, we aim to address these limitations and explore ideas to extend the sequence learning model to deal with real-world tasks such as classifying and detecting anomalies in spiking temporal data.
Your tasks will include:

  • Investigating in online learning methods for sequence processing in the literature
  • Extending the sequence learning model in [3] to permit processing of real-world data, e.g., phoneme classification [5], keyword classification [4]
  • Studying whether the model is robust to noise and perturbation
  • Benchmarking the model on different tasks and comparing results against different state-of-the-art models

Your Profile:

  • Current master studies in physics, computer science, mathematics, electrical/electronic engineering or a related science or engineering field
  • Strong background in mathematics, e.g., probability theory, linear algebra, differential/integral calculus
  • Knowledge of biology and in particular neuroscience is a plus
  • Prior programming experience in Python is a must, C++ and CUDA experience are a plus
  • Hands-on experience in working with neural simulators (NEST, Brian, etc.) and/or machine learning frameworks (PyTorch, Tensorflow, etc.) is a plus
  • Experience with spiking neural networks and/or neuromorphic computing is a plus

Please feel free to apply for the position even if you do not have all the required skills and knowledge. We may be able to teach you missing skills during your introduction.

Our Offer:

We work on the very latest issues that impact our society and are offering you the chance to actively help in shaping the change! We support you in your work with:

  • A world-leading, interdisciplinary and international research environment, provided with state-of-the-art experimental equipment and versatile opportunities
  • Qualified support through your scientific colleagues
  • An interdisciplinary and collaborative work environment including researchers at the following institutes: Neuromorphic Hardware Nodes (PGI-14), Electronics Materials (PGI-7), the Institute of Neuroscience and Medicine - Computational and Systems Neuroscience (INM-6), the Jülich Supercomputer Center (JSC) and the Faculty of Electrical Engineering and Information Technology at RWTH Aachen
  • The chance to independently prepare and work on your tasks
  • Flexible working hours and possibility to work partly remote as well as a reasonable remuneration
  • Working for a distinguished employer in Germany - 6th place in the Glassdoor award for employee satisfaction: https://www.glassdoor.de/Award/Beste-Arbeitgeber-Deutschland-LST_KQ0,29.htm


Place of employment: Aachen

We welcome applications from people with diverse backgrounds, e.g. in terms of age, gender, disability, sexual orientation / identity, and social, ethnic and religious origin. A diverse and inclusive working environment with equal opportunities in which everyone can realize their potential is important to us.

References
[1] Jesus L. Lobo, Javier Del Ser, Albert Bifet, and Nikola Kasabov. Spiking neural networks and online
learning: An overview and perspectives. Neural Networks, 121:88–100, 2020.
[2] A. Mehonic and A. J. Kenyon. Brain-inspired computing needs a master plan. Nature, 604(7905):255–260, April 2022.
[3] Younes Bouhadjar, Dirk J. Wouters, Markus Diesmann, and Tom Tetzlaff. Sequence learning, prediction, and replay in networks of spiking neurons. PLOS Computational Biology, 18(6):e1010233, 2022.
[4] Benjamin Cramer, Yannik Stradmann, Johannes Schemmel, and Friedemann Zenke. The heidelberg spiking data sets for the systematic evaluation of spiking neural networks. IEEE Transactions on Neural Networks and Learning Systems, 33(7):2744–2757, July 2022.
[5] Guillaume Bellec, Darjan Salaj, Anand Subramoney, Robert Legenstein, and Wolfgang Maass. Long short-term memory and learning-to-learn in networks of spiking neurons, 2018.



Similar Positions