Differentially Private Non-Convex Optimization under the KL Condition with Optimal Rates

We study private empirical risk minimization (ERM) problem for losses satisfying the -Kurdyka-{\L}ojasiewicz (KL) condition. The Polyak-{\L}ojasiewicz (PL) condition is a special case of this condition when . Specifically, we study this problem under the constraint of zero-concentrated differential privacy (zCDP). When and the loss function is Lipschitz and smooth over a sufficiently large region, we provide a new algorithm based on variance reduced gradient descent that achieves the rate on the excess empirical risk, where is the dataset size and is the dimension. We further show that this rate is nearly optimal. When and the loss is instead Lipschitz and weakly convex, we show it is possible to achieve the rate with a private implementation of the proximal point method. When the KL parameters are unknown, we provide a novel modification and analysis of the noisy gradient descent algorithm and show that this algorithm achieves a rate of adaptively, which is nearly optimal when . We further show that, without assuming the KL condition, the same gradient descent algorithm can achieve fast convergence to a stationary point when the gradient stays sufficiently large during the run of the algorithm. Specifically, we show that this algorithm can approximate stationary points of Lipschitz, smooth (and possibly nonconvex) objectives with rate as fast as and never worse than . The latter rate matches the best known rate for methods that do not rely on variance reduction.
View on arXiv