A Dimension-free Algorithm for Contextual Continuum-armed Bandits

In contextual continuum-armed bandits, the contexts and the arms are both continuous and drawn from high-dimensional spaces. The payoff function to learn does not have a particular parametric form. The literature has shown that for Lipschitz-continuous functions, the optimal regret is , where and are the dimensions of contexts and arms, and thus suffers from the curse of dimensionality. We develop an algorithm that achieves regret when is globally concave in . The global concavity is a common assumption in many applications. The algorithm is based on stochastic approximation and estimates the gradient information in an online fashion. Our results generate a valuable insight that the curse of dimensionality of the arms can be overcome with some mild structures of the payoff function.
View on arXiv