Summary

• I am performing my PhD at Aix-Marseille Université.
• I am part of the computer science lab LIS and mathematic lab I2M.
• My PhD advisors are Valentin Emiya and Caroline Chaux.

My thesis work consists in using Straight-Through-Estimator (STE) techniques for sparse optimization.
During my thesis, I first focused on developing an algorithm based on straight-through estimators for sparse support recovery. Now, I'm studying sparse matrix factorization. My final research goal is to mitigate the computational expenses of neural networks by reducing the computational cost of matrix-vector products.
I've also joined Yamaha part-time to work on trumpet performance descriptors prediction.