Nathan Buskulic | Machine Learning Center of Genova (MaLGa)

About Me

Hi! I am a postdoctoral researcher in the Machine Learning center of Genova (MaLGa) working mainly with Luca Calatroni. My work focuses on the theoretical aspects of deep learning when used to solve inverse problems. I am particularly interested in the guarantees that one can achieve with deep learning to help apply this technology to critical domains. I obtained my Ph.D. from Université de Caen in Normandy, France, under the supervision of Jalal Fadili and Yvain Quéau.

News

[January 2025] I started my Postdoc at MaLGa (Italy) with Luca Calatroni.
[November 2024] I succesfully obtained my PhD !
[August 2024] Our paper on gradient descent was shortlisted for best paper award at EUSIPCO 2024.
[September 2023] The full journal version of our work on convergence guarantees of DL applied to inverse problems is on arxiv.
[May 2023] Our paper won the best paper award at SSVM.

Selected Publications

Implicit Regularization of the Deep Inverse Prior Trained with Inertia

Nathan Buskulic, Jalal Fadili, Yvain Quéau

Arxiv, 2025

PDF
Recovery Guarantees of Unsupervised Neural Networks for Inverse Problems trained with Gradient Descent

Nathan Buskulic, Jalal Fadili, Yvain Quéau

EUSIPCO, 2024

PDF Finalist for best paper award
Convergence and Recovery Guarantees of Unsupervised Neural Networks for Inverse Problems

Nathan Buskulic, Jalal Fadili, Yvain Quéau

JMIV, 2023

PDF
Convergence Guarantees of Overparametrized Wide Deep Inverse Prior

Nathan Buskulic, Yvain Quéau, Jalal Fadili

SSVM, 2023

PDF Best paper award

Teachings

Université de Caen

Image Synthesis: Ray-tracing methods and creation of a ray-tracer from Nori

Master level, 2022 and 2023, teaching assistant
Image Analysis: Fundamentals of variational methods to solve inverse problems

Master level, 2022 and 2023, teaching assistant