AI RESEARCH
When and why randomised exploration works (in linear bandits)
arXiv CS.LG
•
ArXi:2502.08870v2 Announce Type: replace We provide an approach for the analysis of randomised exploration algorithms like Thompson sampling that does not rely on forced optimism or posterior inflation. With this, we nstrate that in the $d$-dimensional linear bandit setting, when the action space is smooth and strongly convex, randomised exploration algorithms enjoy an $n$-step regret bound of the order $O(d\sqrt{n} \log(n