Fetching the paper…

Efficient Online Bandit Multiclass Learning with $\tilde{O}(\sqrt{T})$ Regret · Around