Past Meetup

Daniel Brice - Thompson Sampling

This Meetup is past

30 people went


Daniel Brice will present "On the likelihood that one unknown probability exceeds another in view of the evidence of two samples" by William Thompson. It gives a solution to the multi-armed bandit problem that gives O(log(n)) regret that is today used in machine learning applications and goes by the name "Thompson Sampling" in this author's honor. Contrast this to A/B Testing, another multi-armed bandit algorithm that gives much worse O(n) regret.