Defensive Boosting for Online Probabilistic Forecasting (arXiv 2608.13554)

Forum topic · 小凯 · 2026-08-15

Summary

Researchers Georgy Noarov and Aaron Roth introduce the Defensive Booster, an online probabilistic forecasting algorithm for binary outcomes chosen by an adaptive adversary. Existing online boosting methods offer incomparable guarantees: online gradient boosting competes in Brier score with the best predictor induced by the span of a weak hypothesis class H but fails when the span lacks an accurate predictor, while online weak-to-strong boosting drives classification error to zero under a weak-learning condition but degrades when it fails. The Defensive Booster achieves both guarantees simultaneously on any adaptive sequence, by operationalizing the dual view of boosting: persistently high randomized classification error yields an ex-post hard-core certificate that the weak-learning condition fails. A strongly adaptive variant satisfies both guarantees on every time interval. The method is highly efficient, accessing only one weak-class learner versus large ensembles in prior approaches, and experiments on synthetic and real data streams show strong predictive performance with orders-of-magnitude faster runtime.

Paper Overview

Field: Machine Learning
Authors: Georgy Noarov, Aaron Roth
Published: 2026-08-13
arXiv: 2608.13554

Introduction

The Defensive Booster

On every adaptive sequence, its Brier score is competitive with the best prediction induced by the span of H at the same rate as online gradient boosting.
Simultaneously, whenever the realized transcript satisfies the smooth weak-learning condition, its Brier score and randomized classification error satisfy the same rate guarantee as online classification boosting.

This is achieved by operationalizing the "dual view" of boosting: when the algorithm's randomized classification error is persistently high, its mistake weights form a smooth reweighting on which every weak hypothesis has low edge, yielding an ex-post hard-core certificate that the weak-learning condition fails.

Strongly Adaptive Variant

We also develop a strongly adaptive variant, which satisfies both guarantees on every time interval.

Efficiency and Experiments

The Defensive Booster is very efficient: it accesses just one weak-class learner, whereas the prior online boosting methods we compare against maintain large weak-learner ensembles. Experiments on synthetic and real data streams demonstrate its strong predictive performance (sometimes substantially improving over all prior baselines) coupled with orders-of-magnitude faster runtime.

--- *Auto-collected on 2026-08-15*