Competing with Markov prediction strategies

Vladimir Vovk

Competing with Markov prediction strategies

Vladimir Vovk

Research output: Working paper

172 Downloads (Pure)

Abstract

Assuming that the loss function is convex in the prediction, we construct a prediction strategy universal for the class of Markov prediction strategies, not necessarily continuous. Allowing randomization, we remove the requirement of convexity.

Original language	English
Number of pages	11
Publication status	Published - 28 Jul 2006

Keywords

machine learning
prediction with expert advice

Access to Document

pdf

Cite this

@techreport{3f27d3262ce94133b25108afcc92a8b3,

title = "Competing with Markov prediction strategies",

abstract = "Assuming that the loss function is convex in the prediction, we construct a prediction strategy universal for the class of Markov prediction strategies, not necessarily continuous. Allowing randomization, we remove the requirement of convexity.",

keywords = "machine learning, prediction with expert advice",

author = "Vladimir Vovk",

note = "Also published in the COLT 2006 proceedings",

year = "2006",

month = jul,

day = "28",

language = "English",

type = "WorkingPaper",

}

TY - UNPB

T1 - Competing with Markov prediction strategies

AU - Vovk, Vladimir

N1 - Also published in the COLT 2006 proceedings

PY - 2006/7/28

Y1 - 2006/7/28

N2 - Assuming that the loss function is convex in the prediction, we construct a prediction strategy universal for the class of Markov prediction strategies, not necessarily continuous. Allowing randomization, we remove the requirement of convexity.

AB - Assuming that the loss function is convex in the prediction, we construct a prediction strategy universal for the class of Markov prediction strategies, not necessarily continuous. Allowing randomization, we remove the requirement of convexity.

KW - machine learning

KW - prediction with expert advice

M3 - Working paper

BT - Competing with Markov prediction strategies

ER -