BISTRO: An Efficient Relaxation-Based Method for Contextual Bandits
This work addresses computational efficiency in contextual bandits for researchers and practitioners, though it appears incremental as it builds on existing relaxation-based methods.
The paper tackles the problem of contextual bandits with i.i.d. covariates and arbitrary rewards by introducing BISTRO, an efficient algorithm that requires d calls to an empirical risk minimization oracle per round, where d is the number of actions, and it uses unlabeled data to simplify computation.
We present efficient algorithms for the problem of contextual bandits with i.i.d. covariates, an arbitrary sequence of rewards, and an arbitrary class of policies. Our algorithm BISTRO requires d calls to the empirical risk minimization (ERM) oracle per round, where d is the number of actions. The method uses unlabeled data to make the problem computationally simple. When the ERM problem itself is computationally hard, we extend the approach by employing multiplicative approximation algorithms for the ERM. The integrality gap of the relaxation only enters in the regret bound rather than the benchmark. Finally, we show that the adversarial version of the contextual bandit problem is learnable (and efficient) whenever the full-information supervised online learning problem has a non-trivial regret guarantee (and efficient).