2 papers
cs.LG2026
When Do Fewer Coordinates Suffice in DP-SGD?
Huiqi Zhang, Fang Xie
Differentially private stochastic gradient descent (DP-SGD) injects noise into every updated coordinate, making the injected noise energy scale with the ambient parameter dimension…
cs.LG2025
AdaDPIGU: Differentially Private SGD with Adaptive Clipping and Importance-Based Gradient Updates for Deep Neural Networks
Huiqi Zhang, Fang Xie
Differential privacy has been proven effective for stochastic gradient descent; however, existing methods often suffer from performance degradation in high-dimensional settings, as…