Security Evaluation

Security evaluation research focuses on assessing the robustness of various systems against attacks, aiming to identify vulnerabilities and improve their resilience. Current efforts concentrate on evaluating the security of deep learning models (including federated and large language models), power systems, and IoT devices, employing techniques like generative adversarial networks (GANs), gradient boosting machines, and multi-task learning frameworks. These analyses utilize diverse metrics, including attack success rates, toxicity scores, and reliability indices, to quantify security posture and inform the development of more secure systems. This work is crucial for mitigating risks associated with increasingly prevalent AI-driven technologies and critical infrastructure.