PrivSGP-VR: Differentially Private Variance-Reduced Stochastic Gradient Push with Tight Utility Bounds.
Zehan ZhuYan HuangXin WangJinming XuPublished in: CoRR (2024)
Keyphrases
- stochastic gradient
- differentially private
- differential privacy
- lower bound
- upper bound
- worst case
- privacy preserving
- stochastic gradient descent
- nearest neighbor classifier
- contingency tables
- data sharing
- search logs
- personal information
- objective function
- privacy preservation
- data sets
- probabilistic model
- support vector