Predictive hints from any stabilizing Luenberger observer make hint residuals uniformly bounded in online least squares, yielding logarithmic regret for nonstochastic prediction despite unbounded trajectories in marginally stable systems.
Dimension-free regret for learning asymmetric linear dynamical systems
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.LG 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
USP paired with VAW delivers O(log^3 T) regret for marginally stable linear dynamical systems with asymmetric hidden matrices.
citing papers explorer
-
Online Nonstochastic Prediction: Logarithmic Regret via Predictive Online Least Squares
Predictive hints from any stabilizing Luenberger observer make hint residuals uniformly bounded in online least squares, yielding logarithmic regret for nonstochastic prediction despite unbounded trajectories in marginally stable systems.
-
The Power of Second Order Methods for Sequence Preconditioning
USP paired with VAW delivers O(log^3 T) regret for marginally stable linear dynamical systems with asymmetric hidden matrices.