Contextual Bandits with Knapsacks for a Conversion Model

TL;DR

This paper introduces a new approach for contextual bandits with knapsacks that models the coupling between rewards and costs through customer conversions, providing regret bounds and adaptable policies for sales with discounts.

Contribution

It proposes a novel structure linking rewards and costs via conversions and develops adaptive policies with regret bounds, extending previous linear models.

Findings

Achieves regret bounds of order (OPT/B) √T.

Develops policies based on upper-confidence estimates.

Extends techniques to non-linear reward-cost structures.

Abstract

We consider contextual bandits with knapsacks, with an underlying structure between rewards generated and cost vectors suffered. We do so motivated by sales with commercial discounts. At each round, given the stochastic i.i.d.\ context $\mathbf{x}_t$ and the arm picked $a_t$ (corresponding, e.g., to a discount level), a customer conversion may be obtained, in which case a reward $r(a,\mathbf{x}_t)$ is gained and vector costs $c(a_t,\mathbf{x}_t)$ are suffered (corresponding, e.g., to losses of earnings). Otherwise, in the absence of a conversion, the reward and costs are null. The reward and costs achieved are thus coupled through the binary variable measuring conversion or the absence thereof. This underlying structure between rewards and costs is different from the linear structures considered by Agrawal and Devanur [2016] (but we show that the techniques introduced in the present article may also be applied to the case of these linear structures). The adaptive policies exhibited solve at each round a linear program based on upper-confidence estimates of the probabilities of conversion given $a$ and $\mathbf{x}$. This kind of policy is most natural and achieves a regret bound of the typical order (OPT/$B$) $\sqrt{T}$, where $B$ is the total budget allowed, OPT is the optimal expected reward achievable by a static policy, and $T$ is the number of rounds.