First year PhD Student in Grenoble.

After a Master Degree in Statistics and Data Science, I started a PhD with Sophie Achard, Julyan Arbel and Guillaume Kon Kam King in the Statify team of Inria Grenoble. My subject focus on graph inference and comparison for fMRI data using bayesian statistics. We are especially interested in quantifying uncertainty during graph analysis.

PhD Subject

Today, many data sets collect information across both space and time. A natural way to model this information is through a graph, where nodes and edges are derived from the observations. This approach is particularly useful in studying brain function, where a node represents a brain region and an edge represents a functional connection between two regions, quantified by methods such as fMRI or EEG.

This graph is usually inferred using first an estimation of the correlation matrix. We propose to work in a bayesian framework that allows us to propagate the uncertainty coming from the correlation matrix to the graph analysis. Overall I focus on several projects that all gather around the idea of evaluating the reliability of graph analysis in realistic set-up.

Interests

I am interested in application in neurosciences, biology and sociology. I focus on interpretability and robustness of models.

Collaborations

Beyond my thesis supervisors, I have the pleasure of collaborating with many other people on various projects. Here are some of them:

One of the first project I worked on consists in the creation of a benchmark for graph estimation in collaboration with Ali Fakhar, Kévin Polisano and Irène Gannaz. Currently a conference paper have been published but we hope to provide a more complete benchmark and an available package for simulating and benchmarking such methods using tools from BenchOpt.

I am currently working on an article with Patrycja Ściślewska to provide an accessible implementation of bayesian uncertainty quantification for graph metrics.