SECUNET

SecuNet is dedicated to advancing the security and robustness of AI models, specifically focusing on tabular data applications in healthcare. Our recent project explores the vulnerabilities of CNN and DNN models to adversarial attacks like PGD, Boundary, Carlini & Wagner, and Brendel & Bethge. By developing effective defense mechanisms such as adversarial training, stochastic distillation, defensive distillation, and ensemble models, we enhance the resilience of AI models in critical areas like cardiovascular disease prediction. At SecuNet, we strive to make AI safer and more reliable for real-world applications, ensuring better healthcare outcomes and robust model integrity