Elias Abad Rocamora

Elias Abad Rocamora is a Researcher and Doctoral Assistant at the École Polytechnique Fédérale de Lausanne (EPFL), affiliated with the Laboratory for Information and Inference Systems (LIONS) under the Institute of Electrical Engineering (IEM) within the School of Engineering (STI). He is concurrently pursuing a Doctoral Program in Electrical Engineering through EPFL's Doctoral School of Engineering (EDoc-EDEE).

His research focuses on adversarial machine learning, robustness certification, and efficient neural network architectures. Key areas include adversarial attacks against vision-language models, certified defenses under bounded perturbations, and regularization techniques to prevent catastrophic overfitting. His work also extends to 3D pointcloud processing using graph-based methods and formal verification of polynomial networks.

Rocamora has contributed to cutting-edge research in AI safety, language model security, and geometric deep learning. His recent work emphasizes dual-domain robustness in multimodal models and efficient single-pass detection mechanisms for LLM jailbreaking inputs. Articles span cybersecurity applications, optimization strategies, and foundational aspects of model robustness.

No scientific awards are currently listed. He collaborates actively within the LIONS lab, focusing on interdisciplinary projects at the intersection of electrical engineering and computer science.