Regret Minimization in Stochastic Non-Convex Learning
via a Proximal-Gradient Approach
Nadav Hallak 1 Panayotis Mertikopoulos 2 Volkan Cevher 3
Abstract problems, and they can adapt to different measures of
regret under different oracle feedback assumptions, e.g.,
This paper develops a methodology for regret
perfect/stochastic gradients or bandit feedback. For
...


雷达卡




京公网安备 11010802022788号







