Neural networks : the official journal of the International Neural Network Society
Adaptive differential privacy mechanism for enhanced deep learning model utility and privacy.
Zhang Xiangfei, Zhang Qingchen
Published: 202510.1016/j.neunet.2025.108345
Abstract
The potential leakage of training data privacy in deep learning has been a topic of concern for researchers and the public. To provide formal and rigorous privacy guarantees, many learning systems integrate differential privacy (DP) with SGD optimize…
Preview only. Read the full abstract at the source