Nantes Université MSCA-PF 2022 - hosting offer at the Electronics and Digital Technologies Institute

Sébastien Le Nours (associated professor) from the Electronics and Digital Technologies Institute (IETR) at Nantes Université is looking for postdoctoral candidates interested in applying for a Marie Skłodowska-Curie Postdoctoral Fellowships (MSCA-PF) in 2022.

Keywords

electronic system-level design, artificial intelligence methods, performance and energy optimized embedded system architectures, multi-processor embedded systems.

What is a MSCA Postdoctoral Fellowship ?

MSCA Postdoctoral fellowships support the career development of researchers for enhancing their individual competence diversification in terms of skills at multidisciplinary level and intersectoral experience. The MSCA-PF grant provides a competitive salary plus mobility and family allowances if applicable, as well as research, training and networking costs. This call to interest concerns the European Postdoctoral Fellowships (12 to 24 months of duration in Europe + optional placement in private sector in Europe of 6 months). For more information : MSCA-PF

Team Description

IETR (labelled by CNRS as UMR 6164) is a research institute specialized in electronic and digital technologies for communication systems (www.ietr.fr/en). This institute covers a large scope of disciplines from analog sensors to high performance digital signal and image processing systems. It offers advanced experimental platforms for the purpose of characterization and proof of concept development. The Communication department where the candidate would be incorporated has a strong expertise in the field of optimizing communication embedded systems under non-functional constraints such as performance, energy, reliability, and security. Nowadays, due to current societal challenges, the integration of high performance artificial intelligence algorithms on embedded devices become of great importance in many application domains. In this context, designing highly optimized hardware and software architectures for such devices requires the definition of innovative methods and tools to favour optimizing such systems.

Profile of the supervisor

Sébastien Le Nours, (https://www.ietr.fr/sebastien-le-nours) is an associate professor at Nantes University specialized in the field of electronic system-level design (ESL) for timing and energy constrained embedded systems. Its research concentrates on the definition of models, methods and tools for efficient performance evaluation of distributed, multi-core systems. Currently, this work is oriented towards the definition of a modelling, simulation-based framework for the optimization of artificial intelligence (AI) architectures under performance and energy constraints. This framework includes fast yet accurate performance models of studied multi-core platforms and a FPGA-based prototyping set-up which allows accurate characterization and validation of the studied systems. The proposed research project will be inserted in the scope of this framework and will contribute to its consolidation.

Research project

Two research axis could be considered by the candidates to build their proposal. The first axis, summarized here as ESL4AI, considers the definition of innovative methods for the optimization of hardware accelerators used for high performance execution of AI algorithms. The challenge concerns here the consideration of data representation optimization (quantization and pruning methods) and data placement in memories and related access methods for the optimization of hardware accelerator architectures. The definition of innovative methods in this field could be considered in conjunction with the framework developed in the research group. The second axis, here called AI4ESL, considers the adoption of AI methods to improve existing ESL design flow and its related steps (performance evaluation, design exploration, model refinement). AI methods based on graph neural networks especially represents a high potential solution to infer optimized architecture models in reasonable time considering user-defined functional and non-functional requirements. The proposal and evaluation of such methods could be considered in conjunction with the tools and use-cases developed in the research group.

What are the eligibility criteria ?

Candidates must hold a PhD at the call deadline (14 Sept. 2022) and have 8 years maximum of full-time research experience after the PhD.

“Mobility rule”: candidates must NOT have resided or carried out their main activity (work, studies, etc.) in France for more than 12 months in the 36 monthsimmediately before the call deadline.

Candidates can be from any nationality.

Although not an eligibility criterion, a good track record is recommended.

 What's in it for me ?

1.Selected candidate will have access to online training programme and one-to-one proposal preparation sessions with our experienced European Officers.

2.Selected candidate will have the opportunity to come to Nantes for two days in the beginning of July to meet his supervisor and to start working on the writing of the project. The grant awarded for this stay will cover travel, hotel and subsistence expenses for 2 days and 2 nights (up to 1000€ for the EU countries and up to 2000€ for the non-EU countries). Nantes Université does not cover the visa fees.

If you are interested and have the required background, apply by sending the application formto Sébastien Le Nours (Sebastien.Le-Nours@univ-nantes.fr) and the European office (europe.drpi@univ-nantes.fr)

More information on the dedicated webpage

• Registration for this scheme does not imply automatic funding
• Applications will be examined in the light of the MSCA PF evaluation criteria, both by your potential supervisor and by the experienced European Officers.
• You should expect feedback for your application by 6 June.
• Selected candidates will be ranked and the short-stay grant will be distributed among them within the limits of the available budget.