Semi-parametric contextual bandits with graph-Laplacian regularization

Non-stationarity is ubiquitous in human behavior and addressing it the contextual bandits challenging. Several works have addressed problem by investigating semi-parametric warned that ignoring non-stationarity could harm performances. Another prevalent social interaction which has become available a form of network or graph structure. As result, graph-based received much attention. In this paper, we propose SemiGraphTS, novel Thompson-sampling algorithm for reward model. Our first to be proposed setting. We derive an upper bound cumulative regret can expressed as multiple factor depending on structure order model without graph. evaluate existing algorithms via simulation real data example.