Universal Algorithms for Probability Forecasting. / Zhdanov, Fedor; Kalnishkan, Yuri.

In: International Journal on Artificial Intelligence Tools, Vol. 21, No. 4, 1240015, 08.2012.

Research output: Contribution to journalArticlepeer-review

Published

Standard

Universal Algorithms for Probability Forecasting. / Zhdanov, Fedor; Kalnishkan, Yuri.

In: International Journal on Artificial Intelligence Tools, Vol. 21, No. 4, 1240015, 08.2012.

Research output: Contribution to journalArticlepeer-review

Harvard

Zhdanov, F & Kalnishkan, Y 2012, 'Universal Algorithms for Probability Forecasting', International Journal on Artificial Intelligence Tools, vol. 21, no. 4, 1240015. https://doi.org/10.1142/S0218213012400155

APA

Zhdanov, F., & Kalnishkan, Y. (2012). Universal Algorithms for Probability Forecasting. International Journal on Artificial Intelligence Tools, 21(4), [1240015]. https://doi.org/10.1142/S0218213012400155

Vancouver

Zhdanov F, Kalnishkan Y. Universal Algorithms for Probability Forecasting. International Journal on Artificial Intelligence Tools. 2012 Aug;21(4). 1240015. https://doi.org/10.1142/S0218213012400155

Author

Zhdanov, Fedor ; Kalnishkan, Yuri. / Universal Algorithms for Probability Forecasting. In: International Journal on Artificial Intelligence Tools. 2012 ; Vol. 21, No. 4.

BibTeX

@article{86957ad9d71648cb82a976875af5b2b2,
title = "Universal Algorithms for Probability Forecasting",
abstract = "Multi-class classification is one of the most important tasks in machine learning. In this paper we consider two online multi-class classification problems: classification by a linear model and by a kernelized model. The quality of predictions is measured by the Brier loss function. We obtain two computationally efficient algorithms for these problems by applying the Aggregating Algorithms to certain pools of experts and prove theoretical guarantees on the losses of these algorithms.We kernelize one of the algorithms and prove theoretical guarantees on its loss. We perform experiments and compare our algorithms with logistic regression.",
author = "Fedor Zhdanov and Yuri Kalnishkan",
year = "2012",
month = aug,
doi = "10.1142/S0218213012400155",
language = "English",
volume = "21",
journal = "International Journal on Artificial Intelligence Tools",
issn = "0218-2130",
publisher = "World Scientific Publishing Co. Pte Ltd",
number = "4",

}

RIS

TY - JOUR

T1 - Universal Algorithms for Probability Forecasting

AU - Zhdanov, Fedor

AU - Kalnishkan, Yuri

PY - 2012/8

Y1 - 2012/8

N2 - Multi-class classification is one of the most important tasks in machine learning. In this paper we consider two online multi-class classification problems: classification by a linear model and by a kernelized model. The quality of predictions is measured by the Brier loss function. We obtain two computationally efficient algorithms for these problems by applying the Aggregating Algorithms to certain pools of experts and prove theoretical guarantees on the losses of these algorithms.We kernelize one of the algorithms and prove theoretical guarantees on its loss. We perform experiments and compare our algorithms with logistic regression.

AB - Multi-class classification is one of the most important tasks in machine learning. In this paper we consider two online multi-class classification problems: classification by a linear model and by a kernelized model. The quality of predictions is measured by the Brier loss function. We obtain two computationally efficient algorithms for these problems by applying the Aggregating Algorithms to certain pools of experts and prove theoretical guarantees on the losses of these algorithms.We kernelize one of the algorithms and prove theoretical guarantees on its loss. We perform experiments and compare our algorithms with logistic regression.

U2 - 10.1142/S0218213012400155

DO - 10.1142/S0218213012400155

M3 - Article

VL - 21

JO - International Journal on Artificial Intelligence Tools

JF - International Journal on Artificial Intelligence Tools

SN - 0218-2130

IS - 4

M1 - 1240015

ER -