Post-Doctoral Research Visit F/M Transfer Learning for Graph-linked Data

The Inria centre at Université Côte d'Azur includes 42 research teams and 9 support services. The centre's staff (about 500 people) is made up of scientists of different nationalities, engineers, technicians and administrative staff. The teams are mainly located on the university campuses of Sophia Antipolis and Nice as well as Montpellier, in close collaboration with research and higher education laboratories and establishments (Université Côte d'Azur, CNRS, INRAE, INSERM ...), but also with the regiona economic players.

With a presence in the fields of computational neuroscience and biology, data science and modeling, software engineering and certification, as well as collaborative robotics, the Inria Centre at Université Côte d'Azur  is a major player in terms of scientific excellence through its results and collaborations at both European and international levels.

The work will be at Inria team NEO (https://team.inria.fr/neo/), located in Sophia Antipolis, under the supervision

of Dr. Konstantin Avrachenkov (https://www-sop.inria.fr/members/Konstantin.Avratchenkov/me.html).

Topic:
The project is a cutting-edge research initiative funded by Université Côte d’Azur under
the DS4H “Idées 2025” program. It aims to push the boundaries of machine learning by
developing advanced techniques for transfer learning in graph-linked data. In many
real-world scenarios—from electrical grids and transportation systems to blockchain
networks—data is naturally represented as graphs. However, most existing Graph Neural
Networks (GNNs) struggle to generalize across different graph topologies, especially
when those graphs are large, dynamic, or only partially labeled. We seeks to overcome
these limitations by creating novel GNN architectures that are invariant (i.e., their
outputs do not change under graph isomorphisms) or equivariant (i.e., their outputs
change in a predictable way when the input graph is transformed). The goal is to
build and to analyse models that maintain high performance even when applied to graphs
that differ from those they were trained on. We shall be testing new methods on
two high-impact use cases: predicting cascading failures in electrical grids and
detecting fraudulent patterns in cryptocurrency networks.

Background references:

Avrachenkov, K., Mishenin, A., Gonçalves, P. & Sokol, M. Generalized optimization framework
for graph-based semi-supervised learning. In Proceedings of SIAM SDM 2012.

Azizian, W., & Lelarge, M. Expressive power of invariant and equivariant graph neural networks.
arXiv preprint arXiv:2006.15646, 2020, also in Proceedings of ICLR 2021.

Gritsenko, A., Shayestehfard, K., Guo, Y., Moharrer, A., Dy, J. & Ioannidis, S.
Graph transfer learning. Knowledge and Information Systems, 65(4), 1627-1656, 2023.

To apply: Submit an application containing CV, Research Statement with a connection to the
proposed project and two recommendation letters.

Main activities:

1. Conducting research;
2. Carrying out numerical experiments (in pyton or matlab);
3. Writing scientific articles

Required skills:

We seek a candidate with PhD in Mathematics or Theoretical Computer Science/Machine Learning.
A candidate should have a solid knowledge in probability, statistics and a good knowledge in
graph theory. A candidate should be able to program in pyton or matlab.

• 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
• Contribution to mutual insurance (subject to conditions)

Gross Salary: 2788 € per month.