Generic Event Prediction and the Big Match (ctd)

28^th November 2018, 13:00
Rasmus Ibsen-Jensen
University of Liverpool

Abstract

I will define the Generic Event Prediction problem (essentially: Predict when an event is going to occour before it does so, as well as possible, without any prior knowledge about the event). All known good strategies for the Big Match can be viewed as solutions to the Generic Event Prediction problem. I will therefore talk a bit about that game and previous and my results about good strategies for it. Last time, I came with a lot of unsubstantiated claims about how to solve the generic event prediction problem I presented. This time I will prove some of them.

These results are joint works with (subsets of) Kristoffer Arnsfelt Hansen, Michal Koucky and Abraham Neyman.
The works appeared in EC'18 and SAGT'16 (Some of my results that have not appeared yet will be mentioned.).