Login / Signup
Output Perturbation for Differentially Private Convex Optimization with Improved Population Loss Bounds, Runtimes and Applications to Private Adversarial Training.
Andrew Lowy
Meisam Razaviyayn
Published in:
CoRR (2021)
Keyphrases
</>
convex optimization
differentially private
differential privacy
privacy guarantees
privacy preserving
total variation
supervised learning
expert advice
multi class
training set
training examples
original data
reinforcement learning
image restoration
model selection
natural images
worst case
optimal solution