The Saddle-Point Accountant for Differential Privacy.
Wael AlghamdiShahab AsoodehFlávio P. CalmonJuan Felipe GómezOliver KosutLalitha SankarFei WeiPublished in: CoRR (2022)
Keyphrases
- saddle point
- differential privacy
- differentially private
- privacy preserving
- variational inequalities
- penalty function
- primal dual
- maximum margin
- privacy preservation
- discrete space
- structured output
- structured prediction
- interior point
- global constraints
- dirichlet distribution
- constrained optimization
- data sharing
- numerical methods
- personal information
- data privacy
- linear programming
- sensitivity analysis
- convex optimization
- search logs
- machine learning
- linear program
- sensitive information