Machine learning & open science
Learning representations
from time series.
I’m Bruno, a research scientist studying how models learn useful representations from time series, with brain signals as a central application.
About me
I work at Yneuro in France and hold an honorary research appointment at UC San Diego. I earned my PhD in Computer Science jointly at Université Paris-Saclay and the Federal University of ABC, advised by Sylvain Chevallier, Marie-Constance Corsi and Raphael Y. de Camargo.
Research interests
My focus is representation learning from time series: learning structure that supports decoding, generation and transfer across subjects, sessions and datasets. I explore these questions through EEG and other neural signals, combining deep learning, Riemannian geometry and reproducible benchmarks.
Open-source software
I build and maintain tools for learning from time series, geometric deep learning and reproducible neuroscience. Alongside these projects, I contribute to MNE-Python, MONAI, MONAI Generative and SpeechBrain.
Mother of All BCI Benchmarks: reproducible evaluation of BCI pipelines across open datasets.
Open data, tools, and compute resource for machine learning on neuroelectromagnetic data.
Geometric (Riemannian/SPD) deep learning library for neural decoding through trivialization.