Online Learning of Smooth Functions

TL;DR

This paper investigates the theoretical limits of online learning for smooth functions, providing sharp bounds and exact results for various function classes and dimensions, advancing understanding of prediction error bounds.

Contribution

It introduces new bounds and exact results for the optimal prediction error in online learning of smooth functions, extending to multi-dimensional cases and different smoothness parameters.

Findings

Sharp bounds for single-variable function classes.

Exact results for specific smoothness and error parameters.

Bounds for multi-variable function classes depending on dimension.

Abstract

In this paper, we study the online learning of real-valued functions where the hidden function is known to have certain smoothness properties. Specifically, for $q \ge 1$, let $\mathcal F_q$ be the class of absolutely continuous functions $f: [0,1] \to \mathbb R$ such that $\|f'\|_q \le 1$. For $q \ge 1$ and $d \in \mathbb Z^+$, let $\mathcal F_{q,d}$ be the class of functions $f: [0,1]^d \to \mathbb R$ such that any function $g: [0,1] \to \mathbb R$ formed by fixing all but one parameter of $f$ is in $\mathcal F_q$. For any class of real-valued functions $\mathcal F$ and $p>0$, let $\text{opt}_p(\mathcal F)$ be the best upper bound on the sum of $p^{\text{th}}$ powers of absolute prediction errors that a learner can guarantee in the worst case. In the single-variable setup, we find new bounds for $\text{opt}_p(\mathcal F_q)$ that are sharp up to a constant factor. We show for all $\varepsilon \in (0, 1)$ that $\text{opt}_{1+\varepsilon}(\mathcal{F}_{\infty}) = \Theta(\varepsilon^{-\frac{1}{2}})$ and $\text{opt}_{1+\varepsilon}(\mathcal{F}_q) = \Theta(\varepsilon^{-\frac{1}{2}})$ for all $q \ge 2$. We also show for $\varepsilon \in (0,1)$ that $\text{opt}_2(\mathcal F_{1+\varepsilon})=\Theta(\varepsilon^{-1})$. In addition, we obtain new exact results by proving that $\text{opt}_p(\mathcal F_q)=1$ for $q \in (1,2)$ and $p \ge 2+\frac{1}{q-1}$. In the multi-variable setup, we establish inequalities relating $\text{opt}_p(\mathcal F_{q,d})$ to $\text{opt}_p(\mathcal F_q)$ and show that $\text{opt}_p(\mathcal F_{\infty,d})$ is infinite when $p<d$ and finite when $p>d$. We also obtain sharp bounds on learning $\mathcal F_{\infty,d}$ for $p < d$ when the number of trials is bounded.