In future articles we’ll discuss expanding upon this framework using more complex Bayesian sampling techniques (in situations with non-conjugate priors), as well as applications to Thompson sampling to new problems: non-binomial multi-armed bandit, path optimization, and more.
