Freezing and Sleeping: Tracking Experts that Learn by Evolving Past Posteriors

Details Freezing and Sleeping: Tracking Experts that Learn by Evolving Past Posteriors Freezing and Sleeping: Tracking Experts that Learn by Evolving Past Posteriors

2010-08-27

arxiv.org/abs/1008.4654v1

Computer Science

Learning

Scientific paper

A problem posed by Freund is how to efficiently track a small pool of experts out of a much larger set. This problem was solved when Bousquet and Warmuth introduced their mixing past posteriors (MPP) algorithm in 2001. In Freund's problem the experts would normally be considered black boxes. However, in this paper we re-examine Freund's problem in case the experts have internal structure that enables them to learn. In this case the problem has two possible interpretations: should the experts learn from all data or only from the subsequence on which they are being tracked? The MPP algorithm solves the first case. Our contribution is to generalise MPP to address the second option. The results we obtain apply to any expert structure that can be formalised using (expert) hidden Markov models. Curiously enough, for our interpretation there are \emph{two} natural reference schemes: freezing and sleeping. For each scheme, we provide an efficient prediction strategy and prove the relevant loss bound.

Affiliated with

Also associated with

No associations

Rating

Freezing and Sleeping: Tracking Experts that Learn by Evolving Past Posteriors does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.

If you have personal experience with Freezing and Sleeping: Tracking Experts that Learn by Evolving Past Posteriors, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Freezing and Sleeping: Tracking Experts that Learn by Evolving Past Posteriors will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFWR-SCP-O-331243