Learning Contextual Bandits Through Perturbed Rewards. (arXiv:2201.09910v1 [cs.LG])

Web: http://arxiv.org/abs/2201.09910

Thanks to the power of representation learning, neural contextual bandit
algorithms demonstrate remarkable performance improvement against their
classical counterparts. But because their exploration has to be performed in
the entire neural network parameter space to obtain nearly optimal regret, the
resulting computational cost is prohibitively high. We perturb the rewards when
updating the neural network to eliminate the need of explicit exploration and
the corresponding computational overhead. We prove that a
$\tilde{O}(\tilde{d}\sqrt{T})$ regret upper bound is still achievable under
standard …

arxiv learning