Filters








2 Hits in 4.4 sec

Do Not Let Privacy Overbill Utility: Gradient Embedding Perturbation for Private Learning [article]

Da Yu, Huishuai Zhang, Wei Chen, Tie-Yan Liu
2021 arXiv   pre-print
However, for meaningful privacy parameters, a differentially private model degrades the utility drastically when the model comprises a large number of trainable parameters.  ...  Then, GEP perturbs the low-dimensional embedding and the residual gradient separately according to the privacy budget.  ...  CONCLUSION In this paper, we propose Gradient Embedding Perturbation (GEP) for learning with differential privacy.  ... 
arXiv:2102.12677v3 fatcat:74l7wqb4fbdjxpkppdpnnaod6u

Differentially Private SGD with Sparse Gradients [article]

Junyi Zhu, Matthew Blaschko
2021 arXiv   pre-print
To protect sensitive training data, differentially private stochastic gradient descent (DP-SGD) has been adopted in deep learning to provide rigorously defined privacy.  ...  However, DP-SGD requires the injection of an amount of noise that scales with the number of gradient dimensions, resulting in large performance drops compared to non-private training.  ...  Test Accuracy Da Yu, Huishuai Zhang, Wei Chen, and Tie-Yan Liu.Do not let privacy overbill utility: Gradient embedding perturbation for private learning.In International Conference on Learning Representations  ... 
arXiv:2112.00845v1 fatcat:g2pduyzclnel7k272sd3g7qgke