Oracle-Efficient and Parameter-Free Agnostic Smoothed Online Learning

Key points

• Online learning is an attractive framework in many domains because it permits well-defined learning even when data are dependent or chosen adversarially.1
• This generality, however, comes at a steep price, introducing significant statistical and computational barriers.1
• Both assumptions limit the applicability of these algorithms, in contrast to statistical learning, where empirical risk minimization (ERM) learns efficiently in the agnostic setting without any knowledge of the data distribution.1
• Our algorithm, based on Gaussian Follow-The-Perturbed-Leader, is parameter-free: it requires no knowledge of $μ$, the smoothing parameter $σ$, or the horizon $T$, and it achieves regret $\widetilde O(d\sqrt{T/σ})$ for binary classes of VC dimension $d$ with a single call to an ERM oracle per round, which is optimal up to a $\sqrt{d}$ factor.1

Sources (1)

• [1]Oracle-Efficient and Parameter-Free Agnostic Smoothed Online Learning
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 7, 05:49 PM

  We show that neither assumption is necessary, giving the first oracle-efficient algorithm that achieves sublinear regret in the agnostic setting without knowledge of $μ$.

  Online learning is an attractive framework in many domains because it permits well-defined learning even when data are dependent or chosen adversarially.

Extractive summary: sentences quoted from the sources.