Moise Beugre

École Nationale Supérieure de l'Électronique et de ses Applications

Papers

Total Citations

H-Index

About

Moise Beugre has established himself as a leading voice in the critical intersection of deep learning security and robotics. His primary research focuses on the vulnerability of neural networks to adversarial attacks, with a particular emphasis on pixel-level perturbations that pose existential threats to autonomous systems. Beugre’s most influential work, "Efficient Detection of Pixel-Level Adversarial Attacks" (2020), has garnered 10 citations and addresses a fundamental challenge: how to safeguard robotic perception systems from subtle, malicious inputs that can completely fool object recognition and scene understanding algorithms. By developing detection methods that identify these nearly invisible alterations, Beugre has contributed essential tools for building more robust and trustworthy AI systems. His research is particularly vital for real-world applications where a single misclassified pixel could lead to catastrophic failures in autonomous vehicles or industrial robots. Through his work, Beugre continues to push the boundaries of adversarial machine learning, ensuring that as deep learning models become more powerful, they also become more resilient against the sophisticated threats that seek to undermine them.