Post-Doctoral Research Visit F/M Post-doctoral Researcher - Expressivity of Implicit neural networks for data-driven stability and safety certificates

The Inria Saclay-Île-de-France Research Centre was established in 2008. It has developed as part of the Saclay site in partnership with Paris-Saclay University and with the Institut Polytechnique de Paris .

The centre has 41 project teams , 27 of which operate jointly with Paris-Saclay University and the Institut Polytechnique de Paris; Its activities occupy over 600 people, scientists and research and innovation support staff, including 44 different nationalities.

This postdoctoral position is in partnership between Inria Saclay, team DISCO, and the Department of Computer Science, University College London, within the UCL-Inria Joint Centre.

The goal of the collaboration is to develop novel methods for computing safety and stability certificates based on neural network parameterizations of Lyapunov functions. Particular attention will be given to analyzing the expressivity of these models as a function of their parameters, including the transition between smooth and non-smooth activation functions. The project will also investigate and compare the expressivity of feedforward and implicit neural networks under equivalent parameterization constraints. In addition, methodological advances will be pursued to improve the training of implicit neural networks.

International collaboration

The job position requires the applicant to spend 12 months in the Inria Saclay, France, and 12 months in London, with a calendar to be defined. 

Mission:
Under the supervision of Prof. Giorgio Valmorbida and Prof. Akin Delibasi, the postdoctoral researcher will be responsible for developing the research program outlined in the research proposal. 

The research proposal is available for download in:

https://filesender.renater.fr/?s=download&token=529793d9-630b-4a96-86c2-478a2ad2bf48

Responsibilities:

The postdoctoral researcher will organize weekly meetings, write monthly progress reports, and give a broader team presentation every 6 months.  As part of a team the researcher will help organise the scientific discussions of the team and, if funding is available, co-supervise MSc students. 

Main activities:

• Develop a study on the expressivity of implicit Neural Networks,
• propose a curriculum learning strategy to transition from smooth activation functions to non-smooth activation functions,
• compute Lyapunov function based on an implicit NN to assess
  • stability and safety,
  • compute control laws based on data-driven control Lyapunov functions. 

Among other activities, the research will 

• Develop NN training strategies for implicit NN. 
• Write documentation of developed software
• Report progress
• Organize project meetings
• Publish the results in the journals and conferences of Automatic Control and Machine Learning
• Communicate the results within the group

Skills: Advanced programming proficiency in Python and MATLAB.

Languages: Fluency in English is required; proficiency in French is considered an asset.

Interpersonal Skills: The candidate will be evaluated on their ability to interact effectively with colleagues and collaborate within an international team. Demonstrated communication skills, including the ability to engage with a broader audience, will be highly valued.

• Subsidized meals
• Partial reimbursement of public transport costs
• Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
• Possibility of teleworking and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities
• Access to vocational training
• Social security coverage

€2,788 gross per month