Differentially Private Coordinate Descent for Composite Empirical Risk Minimization.
Paul MangoldAurélien BelletJoseph SalmonMarc TommasiPublished in: ICML (2022)
Keyphrases
- differentially private
- empirical risk minimization
- differential privacy
- uniform convergence
- statistical learning theory
- empirical risk
- vc dimension
- generalization bounds
- rates of convergence
- machine learning
- privacy preserving
- loss function
- data sets
- multiple kernel learning
- multi task
- decision function
- gaussian kernels
- feature selection
- model selection
- upper bound