Online Nonstochastic Prediction: Logarithmic Regret via Predictive Online Least Squares

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pai, Chih-Fan, Zheng, Yang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913093597003776
author Pai, Chih-Fan
Zheng, Yang
author_facet Pai, Chih-Fan
Zheng, Yang
contents We study online prediction for marginally stable, partially observed linear dynamical systems under nonstochastic disturbances. Our objective is to minimize the cumulative squared prediction loss and compete with the best-in-hindsight Luenberger predictor. Standard online learning methods typically rely on bounded domains/gradients, and thus their guarantees may fail to deal with potentially unbounded trajectories in marginally stable systems. In this paper, we introduce an unconstrained online least squares method that stabilizes the learning process via tailored predictive hints. With model knowledge, we prove that hints constructed from any stabilizing Luenberger predictor render the hint residuals uniformly bounded, achieving logarithmic regret despite unbounded trajectory growth. We also discuss model-free prediction and introduce a simple universal hint for symmetric systems, under which logarithmic regret is maintained without model knowledge. Our results provide an adaptive, instance-wise optimal online predictor compared to classical fixed-gain observers under nonstochastic disturbances.
format Preprint
id arxiv_https___arxiv_org_abs_2605_04364
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Online Nonstochastic Prediction: Logarithmic Regret via Predictive Online Least Squares
Pai, Chih-Fan
Zheng, Yang
Machine Learning
Systems and Control
Optimization and Control
We study online prediction for marginally stable, partially observed linear dynamical systems under nonstochastic disturbances. Our objective is to minimize the cumulative squared prediction loss and compete with the best-in-hindsight Luenberger predictor. Standard online learning methods typically rely on bounded domains/gradients, and thus their guarantees may fail to deal with potentially unbounded trajectories in marginally stable systems. In this paper, we introduce an unconstrained online least squares method that stabilizes the learning process via tailored predictive hints. With model knowledge, we prove that hints constructed from any stabilizing Luenberger predictor render the hint residuals uniformly bounded, achieving logarithmic regret despite unbounded trajectory growth. We also discuss model-free prediction and introduce a simple universal hint for symmetric systems, under which logarithmic regret is maintained without model knowledge. Our results provide an adaptive, instance-wise optimal online predictor compared to classical fixed-gain observers under nonstochastic disturbances.
title Online Nonstochastic Prediction: Logarithmic Regret via Predictive Online Least Squares
topic Machine Learning
Systems and Control
Optimization and Control
url https://arxiv.org/abs/2605.04364