Obfuscation Technique
Obfuscation techniques aim to transform data or code to conceal sensitive information while preserving functionality or utility. Current research focuses on developing and evaluating obfuscation methods for various data types (text, images, code, and even model parameters) using diverse approaches, including large language models, transformer networks, and differential privacy mechanisms. These techniques are crucial for protecting privacy in various applications, from securing sensitive data shared across platforms to mitigating risks associated with adversarial attacks on machine learning models. The ongoing development and rigorous evaluation of these methods are vital for advancing privacy-preserving technologies and ensuring responsible data handling.
Papers
Benchmarking Robustness to Adversarial Image Obfuscations
Florian Stimberg, Ayan Chakrabarti, Chun-Ta Lu, Hussein Hazimeh, Otilia Stretcu, Wei Qiao, Yintao Liu, Merve Kaya, Cyrus Rashtchian, Ariel Fuxman, Mehmet Tek, Sven Gowal
FedPass: Privacy-Preserving Vertical Federated Deep Learning with Adaptive Obfuscation
Hanlin Gu, Jiahuan Luo, Yan Kang, Lixin Fan, Qiang Yang