Uncertainty in Artificial Intelligence
First Name   Last Name   Password   Forgot Password   Log in!
    Proceedings         Authors   Author's Info   Article details         Search    
Vennā??Abers Predictors
Vladimir Vovk, Ivan Petej
Abstract:
This paper continues study, both theoretical and empirical, of the method of Venn prediction, con- centrating on binary prediction problems. Venn predictors produce probability-type predictions for the labels of test objects which are guaran- teed to be well calibrated under the standard as- sumption that the observations are generated in- dependently from the same distribution. We give a simple formalization and proof of this prop- erty. We also introduce Vennā??Abers predictors, a new class of Venn predictors based on the idea of isotonic regression, and report promising empir- ical results both for Vennā??Abers predictors and for their more computationally efficient simpli- fied version.
Keywords:
Pages: 829-838
PS Link:
PDF Link: /papers/14/p829-vovk.pdf
BibTex:
@INPROCEEDINGS{Vovk14,
AUTHOR = "Vladimir Vovk and Ivan Petej",
TITLE = "Vennā??Abers Predictors",
BOOKTITLE = "Proceedings of the Thirtieth Conference Annual Conference on Uncertainty in Artificial Intelligence (UAI-14)",
PUBLISHER = "AUAI Press",
ADDRESS = "Corvallis, Oregon",
YEAR = "2014",
PAGES = "829--838"
}


hosted by DSL   •   site info   •   help