AI RESEARCH
Parameter-Free Dynamic Regret for Unconstrained Linear Bandits
arXiv CS.LG
•
ArXi:2603.25916v1 Announce Type: new We study dynamic regret minimization in unconstrained adversarial linear bandit problems. In this setting, a learner must minimize the cumulative loss relative to an arbitrary sequence of comparators $\boldsymbol{u}_1,\ldots,\boldsymbol{u}_T$ in $\mathbb{R}^d$, but receives only point-evaluation feedback on each round.