Bayesian Optimisation with Unknown Hyperparameters: Regret Bounds Logarithmically Closer to Optimal | Read Paper on Bytez