|
VISION
|
Our vision is to establish a more secure and resilient data environment by enhancing the security of machine learning models through advanced research in attack and defense technologies. By deeply exploring and integrating adversarial attacks with model extraction techniques, we aim to set a new standard for protecting sensitive information processed by AI systems. |
|
MISSION
|
Our mission is to leverage model extraction attacks to bolster the effectiveness of membership inference attacks, and utilize the insights gained from these attacks to train for adversarial attacks. This approach is designed to not only identify vulnerabilities in current models but also to fortify them against potential threats, thereby ensuring a higher level of data protection and integrity across various applications. |
|
TEAM
|
Lin Ouyang, Maohong Liao, Hong Chen |
|
SITE
|
N/A |
