Table 1. Summary of regret for selected algorithms.

Online Learning Algorithm	Closest BCI Algorithm	Regret
Online Gradient Descent	Gradient-based decoder adaptation	$\mathcal{O}\left(\sqrt{K}\right)$ [37] $\mathcal{O}\left(\text{log}K\right)$ *[38]
Moving Average	SmoothBatch [13]	$\mathcal{O}\left(K\right)$
Follow-the-leader	CLDA-style maximum likelihood [14]	$\mathcal{O}(logK)$ [38]

* Bound obtained only under restrictive conditions (see main text).