Adaptive Online Estimation of Piecewise Polynomial Trends

TL;DR

This paper introduces an adaptive online algorithm for estimating piecewise polynomial trends in non-stationary stochastic optimization, achieving nearly optimal dynamic regret and applicable to various non-parametric settings.

Contribution

It proposes a new variational constraint for piecewise polynomial comparators and develops a polynomial-time algorithm with adaptive, nearly minimax optimal regret bounds.

Findings

Achieves nearly optimal dynamic regret of n^{1/(2k+3)} C_n^{2/(2k+3)}

Algorithm is adaptive to unknown variation radius C_n

Proven minimax optimal for multiple non-parametric families

Abstract

We consider the framework of non-stationary stochastic optimization [Besbes et al, 2015] with squared error losses and noisy gradient feedback where the dynamic regret of an online learner against a time varying comparator sequence is studied. Motivated from the theory of non-parametric regression, we introduce a new variational constraint that enforces the comparator sequence to belong to a discrete $k^{th}$ order Total Variation ball of radius $C_n$. This variational constraint models comparators that have piece-wise polynomial structure which has many relevant practical applications [Tibshirani, 2014]. By establishing connections to the theory of wavelet based non-parametric regression, we design a polynomial time algorithm that achieves the nearly optimal dynamic regret of $\tilde{O}(n^{\frac{1}{2k+3}}C_n^{\frac{2}{2k+3}})$. The proposed policy is adaptive to the unknown radius $C_n$. Further, we show that the same policy is minimax optimal for several other non-parametric families of interest.