Ambra Demontis
Papers
1
Total Citations
77
H-Index
1
About
Ambra Demontis is a leading researcher in the security and robustness of machine learning, with a particular focus on adversarial machine learning and its implications for embodied AI systems. Her work has been instrumental in revealing critical vulnerabilities in deep neural networks when deployed in real-world, safety-critical applications. Her most-cited paper, “Is Deep Learning Safe for Robot Vision? Adversarial Examples against the iCub Humanoid” (2018, 77 citations), demonstrated that even physically-realizable adversarial perturbations can fool state-of-the-art vision systems on humanoid robots, raising urgent questions about the safety of autonomous agents. Beyond this, Demontis has made foundational contributions to understanding how adversarial examples transfer across models and how to design more robust learning algorithms. Her research bridges the gap between theoretical security analysis and practical deployment, earning her recognition as a key voice in trustworthy AI. Her work continues to shape how researchers and engineers approach the safety of neural networks in high-stakes environments, from robotics to autonomous driving.
Research Focus
Key Achievements
Top Papers
- 1