Contextual Bandits with Cross-learning

In the classical contextual bandits problem, in each round $t$, a learner observes some context $c$, chooses some action $i$ to perform, and receives some reward $r{i,t}(c)$. We consider the variant of this problem where in addition to receiving the reward $r{i,t}(c)$, the learner also learns the values of $r{i,t}(c')$ for some other contexts $c'$ in set $\mathcal{O}i(c)$; i.e., the rewards that would have been achieved by performing that action under different contexts $c'\in \mathcal{O}i(c)$. This variant arises in several strategic settings, such as learning how to bid in non-truthful repeated auctions, which has gained a lot of attention lately as many platforms have switched to running first-price auctions. We call this problem the contextual bandits problem with cross-learning. The best algorithms for the classical contextual bandits problem achieve $\tilde{O}(\sqrt{CKT})$ regret against all stationary policies, where $C$ is the number of contexts, $K$ the number of actions, and $T$ the number of rounds. We design and analyze new algorithms for the contextual bandits problem with cross-learning and show that their regret has better dependence on the number of contexts. Under complete cross-learning where the rewards for all contexts are learned when choosing an action, i.e., set $\mathcal{O}i(c)$ contains all contexts, we show that our algorithms achieve regret $\tilde{O}(\sqrt{KT})$, removing the dependence on $C$. For any other cases, i.e., under partial cross-learning where $|\mathcal{O}i(c)|< C$ for some context-action pair of $(i,c)$, the regret bounds depend on how the sets $\mathcal Oi(c)$ impact the degree to which cross-learning between contexts is possible. We simulate our algorithms on real auction data from an ad exchange running first-price auctions and show that they outperform traditional contextual bandit algorithms.