Dylan Sechet

CS PhD Student • LISN, Université Paris-Saclay

Dylan Sechet profile photo

Hi! I am a PhD student working in the A&O team at LISN, under the supervision of Matthieu Kowalski and Marc Evrard. Previously, I was working as a research engineer on M/EEG inverse problems.

I received my Master's degree from CentraleSupélec.

Research

My work focuses on building lightweight and explainable neural network for music source separation. I'm exploring topics such as unrolling, time-frequency decompositions and matrix factorization of audio signals.

News

  • 03/25: My paper on Sparse Bayesian Learning has been accepted at SampTA 2025 in Vienna.
  • 09/24: I was in Guildford to present my work on Hierarchical Models for instrument tagging at DAFx24.
  • 08/24: I've just started work at LISN as a research engineer. I'll be building a toolbox for inverse problems in M/EEG.

Selected Publications

Revisiting CHAMPAGNE: Sparse Bayesian Learning as Reweighted Sparse Coding thumbnail
Revisiting CHAMPAGNE: Sparse Bayesian Learning as Reweighted Sparse Coding
Dylan Sechet, Matthieu Kowalski, Samy Mokhtari, Bruno Torresani
International Conference on Sampling Theory and Applications (SampTA), 2025
A Hierarchical Deep Learning Approach for Minority Instrument Detection thumbnail
A Hierarchical Deep Learning Approach for Minority Instrument Detection
Dylan Sechet, Francesca Bugiotti, Matthieu Kowalski, Edouard d'Hérouville, Filip Langiewicz
International Conference on Digital Audio Effects (DAFx), 2024