PostDoc position on causal discovery @ Inria&Nokia Bell Labs

[Eric Gaussier]

Inria and Nokia Bell Labs offer a POstDoc position on causal discovery of extended summary causal graphs for noisy-OR models of event sequences starting as soon as possible and no later than October or November 2023.

Goals The objective of this project is to develop methods to infer causal graphs from observational time series/event-type data generated according to generic noisy-OR models [1]. The causal graphs considered can either be full window causal graphs or a summarized version as extended summary causal graphs [2] and may contain or not hidden common causes. Generic noisy-OR models are structural causal models (SCM) with noisy-OR gates which allow to estimate the effect of multiple causes even if they have never been observed together. We will consider here both simple noisy-OR models in which the noisy-OR gates directly define the SCM, and complex ones in which the noisy-OR gates are sub-parts of an underlying SCM. 

To apply Interested candidates should send a complete CV with a list of publications and two reference letters to Armen Aghasaryan armen.ag...@nokia-bell-labs.com, Emilie Devijver emilie....@univ-grenoble-alpes.fr, Eric Gaussier eric.g...@imag.fr and Gregor Goessler gregor....@inria.fr. Candidates should be pursuing internationally recognized research in ML/AI, or Information Theory with a strong interest in causal inference and causal reasoning.

 

References 

[1] L. Jakovljevic, D. Kostadinov, A. Aghasaryan, T. Palpanas. Towards building a digital twin of complex systems using causal modelling. Complex Networks, 2021.

[2] C. K. Assaad, E. Devijver, E. Gaussier. Discovery of extended summary graphs in time series. Uncertainty in Artificial Intelligence, 2022.

Additional information can be found at https://ama.liglab.fr/~gaussier/JobOffers/Nokia-PostDoc.pdf

--

Eric Gaussier
Grenoble University, LIG Laboratory