PhD position Multiscale information integration in neuromorphic computing

Research / Academic
This position is funded through the European Horizon 2020 project "Memory technologies with multi-scale time constants for neuromorphic architectures" (MeM-Scales). The overall objectives in this project is the development of novel neural-network oriented microchip technologies (analog, spiking, memristor-based) and novel computational techniques which can cope with the brain-like properties of such unconventional hardware. The MeM-Scales project is a follow-up to the very successful project "Neural computing architectures in advanced monolithic 3D-VLSI nano-technologies" (NeuRAM3, The project consortium is broadly interdisciplinary, joining 9 partnering research groups from material science through microchip design & manufacturing to computational neuroscience and machine learning.

Within this context, the focus theme of the MeM-Scales project is to tackle the challenge of multiple timescales. Biological brains can integrate information across timescales that span many orders of magnitude, from millisecond visual information processing to many years of long-term memory consolidation and forgetting. Biological brains employ a large number of physiological processes as a physical basis for operating on a continuum of timescales. On digital microchips used in online information processing, timescales come for free because bits of information can be stored in memory registers for arbitrarily long timespans. However, in future non-digital neuromorphic microchips, reliable non-volatile bit memories will not be available (just as they are not available in biological brains). Instead, dynamical / algorithmical mechanisms for timescale management and information stabilisation must be found and exploited.

Within the MeM-Scales project, the group of Herbert Jaeger ( in Groningen will coordinate the conceptual, mathematical and algorithmical sides of such timescale management. The scientific mission statement for this PhD position could thus be declared as "explore temporal information integration over multiple timescales in unconventional neuromorphic microchips". The general research attitude in Jaeger's group is one of a close coupling between formal mathematical modeling and the design of efficient algorithmic procedures. Since at this point, numerous but hardly explored options are "in the air" for timescale management, this position does not prescribe a specific set of tasks, but offers much freedom to engage in original work.


The successful PhD candidate should have a master degree in computer science, AI, machine learning, signal processing and control, computational neuroscience, theoretical physics or mathematics (dynamical systems) or related fields.

The three main selection criteria are:
- the ability and inclination for rigorous conceptual thinking and mathematical formalization
- a strong interest in solving the riddles of information processing in complex dynamical systems (whether they be neural, technological, or even social)
- openness for working in a very interdisciplinary setting.

Salary Benefits:

We offer you in accordance with the Collective Labour Agreement for Dutch Universities:

• a salary of € 2,325 gross per month in the first year, up to a maximum of € 2,972 gross per month in the fourth and final year, based on a full-time position (1.0 FTE)
• a holiday allowance of 8% gross annual income
• an 8.3% year-end bonus
• a position for four years; you will get a temporary position of one year with the option of renewal for another three years; prolongation of the contract is contingent on sufficient progress in the first year to indicate that a successful completion of the PhD thesis within the next three years is to be expected
• a PhD training programme is part of the agreement and the successful candidate will be enrolled in the Graduate School of Science and Engineering.
Work Hours:

38 hours per week


Broerstraat 5