Faster online calibration without randomization: interval forecasts and the power of two choices

TL;DR

This paper demonstrates that using a small interval forecast with the power of two choices allows for faster calibration rates in adversarial settings without the need for randomization.

Contribution

It introduces a novel approach using interval forecasts and the power of two choices to achieve faster calibration rates in online learning.

Findings

Achieves an $O(1/T)$ calibration rate without randomization.

Interval forecasts with two choices provide significant calibration advantages.

Faster calibration is possible even against adaptive adversaries.

Abstract

We study the problem of making calibrated probabilistic forecasts for a binary sequence generated by an adversarial nature. Following the seminal paper of Foster and Vohra (1998), nature is often modeled as an adaptive adversary who sees all activity of the forecaster except the randomization that the forecaster may deploy. A number of papers have proposed randomized forecasting strategies that achieve an $\epsilon$-calibration error rate of $O(1/\sqrt{T})$, which we prove is tight in general. On the other hand, it is well known that it is not possible to be calibrated without randomization, or if nature also sees the forecaster's randomization; in both cases the calibration error could be $\Omega(1)$. Inspired by the equally seminal works on the "power of two choices" and imprecise probability theory, we study a small variant of the standard online calibration problem. The adversary gives the forecaster the option of making two nearby probabilistic forecasts, or equivalently an interval forecast of small width, and the endpoint closest to the revealed outcome is used to judge calibration. This power of two choices, or imprecise forecast, accords the forecaster with significant power -- we show that a faster $\epsilon$-calibration rate of $O(1/T)$ can be achieved even without deploying any randomization.