When to spend privacy: optimal allocation in private learning
Privacy accounting bounds the cumulative privacy cost of Differentially Private Stochastic Gradient Descent but does not determine how to allocate a fixed budget across iterations. We formulate the complete Gaussian noise-variance schedule as an open-loop control under a composed zero-Concentrated Differential Privacy constraint, explicitly accounting for schedule selection. For expected…
We haven't written up this one. arXiv cs.AI has the full story — the link below goes straight to it.
