ConceptFTRL
Follow-the-Regularized-Leader
A commonly used algorithmic framework in online optimization that selects each round's decision by minimizing the cumulative loss plus a regularization term, widely applied in online learning and online convex optimization.
Timeline (last 90 days)
Oct 22
The algorithm is based on the differentially private Follow-the-Regularized-Leader framework, adaptively adjusting the learning rate and noise scale to exploit the concentration of the loss sequence when L* is small in order to reduce the impact of noise.
Unverified50%