Student project: Efficient Mapping of Spiking Neural Networks into imec's
large-scale neuromorphic processor
What you will do
In the context of two European research projects (TEMPO/ANDANTE), the imec
neuromorphic team is building a scalable neuromorphic processor. Our
neuromorphic processor contains many RISC-V cores connected through an
interconnected network. One of the challenges in this project is to
efficiently map a large-scale neural network into our homogeneous multicore
processor. A mapping algorithm defines how to allocate the available resources
in each core (memory and processing time) to efficiently distribute the load
and therefore reduce the energy and latency of processing. Additionally, it
should reduce the distance between the interconnected cores to minimize the
data movement among the cores 12. Compared to the previous works, the
extra flexibility of our processor intensifies the importance of an efficient
The project duration is 12 months. This project should result in a mapping
tool/algorithm that generates efficient mapping topologies for given neural
networks. The outcome of the project may be published in a high-impact
The application deadline is 31/July/2021. Your application should contain
an up-to-date CV, transcripts for both bachelor's and master's courses,
and optionally a recommendation letter. A motivation letter and statement of
purpose are not mandatory. After the deadline, the short-listed students will
be interviewed by the supervisors and human resource management. The project
shall start in summer 2021. Please be advised that non-EU/EEA country students
that are studying outside of the Netherlands, need to have a work permit to be
able to do an internship in the Netherlands.
1 A. Balaji et al., "Mapping Spiking Neural Networks to Neuromorphic
Hardware," in IEEE Transactions on Very Large Scale Integration (VLSI)
Systems, vol. 28, no. 1, pp. 76-86, Jan. 2020, doi:
2 Mapping Deep Neural Networks on SpiNNaker2
(https: // www. youtube.com/watch?v=fjG0zKGVfw).
Literature review on state-of-the-art mapping optimization work and
Finding/designing a suitable algorithm for mapping SNN into imec
Implementation of the mapping tool in software (for example Python).
Validating the tool by applying it on several large-scale neural networks.
Paper writing and documentation in Imec Holst-Centre.
What we do for you
Imec will provide highly engaged project supervision with experts in the
neuromorphic domain, compensation for expenses and a desk in the Holst Centre.
Who you are
You are an MSc student in Mathematics, Software engineering, or relevant
You are available for 12 months.
Very good/excellent analytical skills (especially in optimization).
Very good/excellent programming skills (preferably in python).
Being familiar with neural networks and neuromorphic systems is a plus.
Entitled to do an internship in the Netherlands.
Motivated student and eager to work independently and expand knowledge in
Good written and verbal English skills.
Click on ‘apply' to submit your application. You will then be redirected to
Please be advised that non-EU/EEA country students that are studying outside
of the Netherlands, need to have a work-permit to be able to do an internship
in the Netherlands.