Geovani Rizk

Biography

I am a postdoctoral researcher at École Polytechnique Fédérale de Lausanne since 2022, where I work with Rachid Guerraoui on robustness of distributed machine learning algorithms. From 2018 to 2022, I completed my Ph.D. in computer science at Université Paris Dauphine-PSL on multi-agent bandits in graphs under the supervision of Yann Chevaleyre, Rida Laraki, Igor Colin, and Albert Thomas.

Publications

Byzantine-Robust Federated Learning: Impact of Client Subsampling and Local Updates
Y Allouah, S Farhadkhani, R Guerraoui, N Gupta, R Pinot, G Rizk, S Voitovych – ICML (2024) 

Robust Distributed Learning: Tight Error Bounds and Breakdown Point under Data Heterogeneity
Y Allouah, R Guerraoui, N Gupta, R Pinot, G. Rizk – NeurIPS (2023)

Stochastic Graphical Bilinear Bandits
G Rizk – Ph.D. Thesis (2022)

An α-No-Regret Algorithm For Graphical Bilinear Bandits.
G Rizk, A Thomas, I Colin, R Laraki, Y Chevaleyre – NeurIPS (2022)

Best Arm Identification In Graphical Bilinear Bandits.
G Rizk, A Thomas, I Colin, R Laraki, Y Chevaleyre – ICML (2021)

Randomization Matters. How To Defend Against Strong Adversarial Attacks.
R Pinot, R Ettedgui, G Rizk, Y Chevaleyre, J Atif – ICML (2020)

Refined Bounds For Randomized Experimental Design.
G Rizk, A Thomas, I Colin, M Draief – NeurIPS Workshop, ML with Guarantees (2019)

Pre-prints

Overcoming the Challenges of Batch Normalization in Federated Learning
R Guerraoui, R Pinot, G Rizk, J Stephan, F Taiani – Preprint, Under review (2024)

Boosting Robustness by Clipping Gradients in Distributed Learning
Y Allouah, R Guerraoui, N Gupta, A Jellouli, G Rizk, J Stephan – Preprint, Under review (2024)