3
2
1
Postdoctorant en informatique spécialisé en analyse des erreurs numériques dans les réseaux de neurones artificiels.
|
|
Deterministic and probabilistic rounding error analysis of neural networks in floating-point arithmeticIMA Journal of Numerical Analysis, In press |
Method for training an artificial intelligence model and associated computer programUnited States, Patent n° : US20250232224A1. 2025 |
|
Method, computer program and device for quantizing a deep neural networkUnited States, Patent n° : US 2023/0334301. 2023 |
|
|
Backward error analysis of artificial neural networks with applications to floating-point computations and adversarial attacksComputer Science [cs]. Université de Toulouse, 2024. English. ⟨NNT : 2024TLSEP054⟩ |
|
|
Adversarial attacks via Sequential Quadratic Programming2022 |
|
|
Adversarial attacks via backward error analysis2021 |