Missions

This project aims to develop a new generative framework for modelling the neural circuitry of the neuronal networks of small animals. The framework relies on generative network models based on embedding the nodes in a dual metric space where distances determine link probabilities based on a kernel that is learned from data. The project will involve generalizing latent space graph models by learning the distance kernels defining the connection probabilities between neurons from data and possibly extending them to non-euclidean embedding spaces.

Activités

- Develop Python code for efficient inference of latent node positions based on stochastic optimization of the data likelihood.
- Development of a procedure for selection of the embedding dimension based on cross validation or Bayesian model selection.
- Possibly extend the model to non-Euclidian embedding spaces.

Compétences

We are looking for a highly motivated candidate with a strong quantitative background in either physics or applied mathematics (including but not limited to machine learning and statistics). Fluency in Python and numerical simulations is expected.

Contexte de travail

We rely in the lab mainly on the Drosophila melanogaster larva as model animal to study these questions. Its full connectome, containing ~12,000 neurons, has been mapped at synaptic resolution. Furthermore, a vast genetics toolbox makes it possible to target and control individual neurons in freely behaving animals. Large-scale screens have revealed the individual influence of thousands of neurons on the behavior in millions of larvae, and several microcircuits controlling specific behavioral decisions and actions have been identified

The selected candidate will work in a highly interdisciplinary environment mixing physicists, biologists, and mathematicians.