Enlarge the image
©Université de Paris

From

Timetable to

Place Campus Saint-Germain-des-Prés, U. Paris Cité

Conferences & seminars

Borelli@Saints-Pères : C. Fettal & A. Chavanne

03/04/23 :
Chakib Fettal: Simple clustering of graphs.
Alice Chavanne: Anxiety onset in adolescents: a machine-learning study.
Room Rabelais 2

Add to the calendar

Chakib Fettal and Alice Chavanne present their work at the Borelli@Saints-Pères seminar.

Chakib Fettal

Title: Simple clustering of graphs

Abstract:  Graph clustering is an important technique used to identify groups of similar nodes within a graph, revealing underlying patterns and structures that may not be immediately apparent. The thesis aims to address scalability issues of the current graph clustering models and presents approaches for clustering and representation learning of different types of graphs using simple techniques. The proposed methods are competitive with state-of-the-art methods while being more computationally efficient, as demonstrated through extensive experimentation and significance testing.

Alice Chavanne

Title: Anxiety onset in adolescents : a machine-learning study

Abstract: Anxiety disorders are among the most prevalent psychiatric disorders, and their onset occurs during adolescence in more than 50% of cases. Neuroimaging MRI approaches have identified brain regions involved in clinical anxiety, however, with small sample sizes and group-level statistics, these findings have struggled to reach clinical utility. As such, this project focused on the analysis of adolescent and young adult neuroimaging data from a large international consortium dataset. Supervised machine-learning techniques, which allow for individual-level prediction, were used to examine the predictive value of psychometric and neuroimaging data in the development of future anxiety disorders. In adolescents from the community, self-report questionnaire scores such as neuroticism showed moderate predictive value for the onset of any anxiety disorder 4 to 8 years later (AUC = 0.68), and neurostructural MRI data contributed to the prediction of a specific anxiety disorder. These results are an additional step towards the understanding of anxiety in the brain, as well as the possibility of early detection of anxiety disorders.