Improved regret analysis for variance-adaptive linear bandits and horizon-free linear mixture MDPs, Yeoneung Kim, Insoon Yang, and Kwang-Sung Jun Advances in Neural Information Processing Systems (NeurIPS), 2022.