2022-05250 - Harnessing Simulation-Based Inference and Lifted Inference to Solve Large-Scale Hierarchical Bayesian Models with Applications to Neurosurgery
Le descriptif de l’offre ci-dessous est en Anglais

Contrat renouvelable : Oui

Niveau de diplôme exigé : Bac + 4 ou équivalent

Fonction : Stagiaire de la recherche

Contexte et atouts du poste

Within the framework of a partnership 

  • collaboration ({team_Inria}, the Lariboisiere Hospital, and the Paris Descartes University

The recruited candidate is espected to develop a mathematical model and a software prototype for the application of simulation-based inference techonolgies to the neurosurgical scenarios.


Large scale probabilistic hierarchical Bayesian models (HBMs) have been successfully used in several research disciplines, including neuroscience. Lately, these have been particularly leveraged to capture population as well as individual effects in neuroimaging, namely image-based phenotyping. Such phenotyping techniques have applications to elucidate relationship between brain structure and neuropathology. Nonetheless, two main obstacles are hampering the generalized use of HBMs in fields such as neuroscience. First, the large dimensionality of these models hinders the use of classical sampling-based approaches such as MCMC and its variants. This leads to the second obstacle, which is that models are constrained to using hierarchical models where efficient algorithms, such as expectation maximization (EM) can be derived, thus restricting the probabilistic laws used for the hierarchical models developed and the applicability of such techniques. The core of this project is to harness the combined power of simulation- based inference and lifted inference to overcame these two obstacles and develop a deep-learning based system where, from the declaration of a parameterized model, and a large dataset, we can fit the dataset and obtain full posterior distributions for model parameters. We will apply this to cognitive neuroscience and neurosurgical planning problems.

Mission confiée

Assignments :
With the help of D. Wassermann the recruited person will be taken to produce a deep-learning based system where, from the declaration of a parameterized forward model, and a large dataset, we can fit the dataset and obtain full posterior distributions for model parameters and apply this to the prediction of cognitive abilities from neuroimaging data.

This project is based on our first proposals, ADAVI and PAVI, where we leverage the structure of pyramidal hierarchical models to produce a deep-learning system to harness simulation-based inference order to fit large scale HBMs.

For a better knowledge of the proposed research subject :

we recommend reading Rouillard, L. & Wassermann, D. ADAVI: Automatic Dual Amortized Variational Inference Applied To Pyramidal Bayesian Models. in ICLR (2022) and the references within.

Responsibilities :

The person recruited is responsible for developing and implementing new HBM-based large scale models in a simulation based inference approach on large and medium sized databases and will take initiatives for reduce the data requirements for model training.


Principales activités

Main activities (5 maximum) :

  • Analyse the requirements of the neuroimaging and neurosurgical applications
  • Develop pipelines for the application of ADAVI-based models to the neuroimaging and neurosurgical cases
  • Design an experimental platform for the implementation of the above pipelines
  • Validate the advancements and write scientific literature on them



Technical skills and level required :

  • Good master of python programming
  • Comfortable with math formalisms and the formal background of machine learning and AI
  • Skills in development deep learning models are desirable

Languages :

  • The candidate is expected to be able to communicate profficiently in English



  • 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 (after 6 months of employment) 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