Reinforcement Learning (RL) is a versatile and powerful tool for control, but often data-inefficient, requiring numerous updates and non-local information such as replay buffers and batch-based processing. This project will investigate event-driven learning approaches in the context of RL in an event-triggered fashion. Data efficiency will be improved by using meta-learning and pre-training that facilitate few-shot adaptations. The research will show how the complementarity of event-triggered learning and meta-learning can drastically increase RL efficiency at test time.
The co-design of algorithms and digital neuromorphic hardware is an additional avenue for enhancing the efficiency of the methods. In this context the research will explore digital, event-based implementations of the event-based RL learning rule. Benchmarking criteria include accuracy, latency, data efficiency, and energy consumption to reach a learned solution on small robotic control tasks. Your tasks in detail:
- Develop an event-driven RL algorithm that sparsely updates network state and parameters that will significantly improve energy to-solution efficiency compared to conventional digital accelerators when applied to control problems or tiny RL scenarios.
- Explore digital hardware realizations of the proposed RL algorithms within existing spiking neural network chip designs.
- Quantitative comparisons with different hardware backends. Design conventional (GPU-based) deep neural networks for comparison.
- Publish research articles, regular participation in top international conferences to present your work.
- Complete two 6-month internships at TU Delft (Prof. Charlotte Frenkel) and at Mercedes-Benz, Böblingen.
- Participate in yearly retreats organized by the doctoral network participants
- Support the dissemination of software tools and concepts.
- Supervise student projects and BSc/MSc theses.
Table of Content
Summary
Subscribe for Scholarship Alert!
Benefits
- The prestige and career benefits of being a Marie Sk?odowska-Curie Actions (MSCA) Doctoral Fellow, including international visibility, research excellence, mobility opportunities, and competitive salary arrangements.
- A world-leading, interdisciplinary and international research environment, provided with state-of-the-art experimental equipment and versatile opportunities to grow as a curious researcher.
- 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 Supercomputing Center (JSC) and the Faculty of Electrical Engineering and Information Technology at RWTH Aachen.
- 30 days of annual leave and provision for days off between public holidays and weekends (e.g. between Christmas and New Year)
- Flexible working hours
- Further development of your personal strengths, e.g. through an extensive range of training courses; a structured program of continuing education and networking opportunities specifically for doctoral researchers via JuDocS, the Jülich Center for Doctoral Researchers and Supervisors: https://www.fz-juelich.de/en/judocs
- You will be enrolled in the PhD program of the department of Electrical Engineering and Information Technology, RWTH Aachen.
- Targeted services for international employees, e.g. through our International Advisory Service
Requirements
- Master’s degree in physics, electrical/electronic engineering, computer science, mathematics, or a related field.
- Strong coding skills for programming neural networks, machine learning and machine learning software frameworks (e.g. PyTorch or Jax) is a must.
- The ability for creative and analytical thinking across discipline boundaries and abstraction levels is a must.
- Knowledge in integrated circuit design, testing and simulation using Cadence is a plus.
- Knowledge of digital neuromorphic hardware and sensors is a plus.
- Ability for collaborative work, interdisciplinary and cross-topical thinking.
- Very good communication skills in English, both spoken and written. PGI-15 offers an English speaking environment, therefore German language skills are not required.
Application Deadline
September 19, 2025How To Apply
Are you qualified and interested in this opportunity? Kindly go to
Forschungszentrum Jülich on www.fz-juelich.de to apply
For more information, kindly visit Forschungszentrum Jülich scholarship webpage.
