Optimal Rates for Random Order Online Optimization

06/29/2021
by   Uri Sherman, et al.
0

We study online convex optimization in the random order model, recently proposed by <cit.>, where the loss functions may be chosen by an adversary, but are then presented to the online algorithm in a uniformly random order. Focusing on the scenario where the cumulative loss function is (strongly) convex, yet individual loss functions are smooth but might be non-convex, we give algorithms that achieve the optimal bounds and significantly outperform the results of <cit.>, completely removing the dimension dependence and improving their scaling with respect to the strong convexity parameter. Our analysis relies on novel connections between algorithmic stability and generalization for sampling without-replacement analogous to those studied in the with-replacement i.i.d. setting, as well as on a refined average stability analysis of stochastic gradient descent.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset