The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Um dos problemas mais importantes no aprendizado de máquina é prever um valor binário observando uma sequência de resultados, até o intervalo de tempo atual, gerados a partir de alguma fonte desconhecida. Vovk e Cesa-Bianchi et al. propuseram de forma independente um modelo de previsão on-line onde se presume que os algoritmos de previsão recebem uma coleção de estratégias de previsão chamadas especialistas e, portanto, podem usar as previsões que fazem. Neste modelo, nenhuma suposição é feita sobre a forma como a sequência de bits a ser prevista é gerada, e o desempenho do algoritmo é medido pela diferença entre o número de erros que ele comete na sequência de bits e o número de erros cometidos pelo algoritmo. o melhor especialista na mesma sequência. Neste artigo estendemos o modelo introduzindo uma noção de investimento. Ou seja, tanto o algoritmo de previsão como os especialistas são obrigados a fazer apostas nas suas previsões em cada intervalo de tempo, e o desempenho do algoritmo é agora medido em relação ao dinheiro total perdido, e não ao número de erros. Analisamos os nossos algoritmos na situação particular em que todos os especialistas partilham a mesma quantidade de apostas em cada intervalo de tempo. Neste modelo de aposta compartilhada, fornecemos um algoritmo de previsão que é, em certo sentido, ideal, mas impraticável, e também fornecemos um algoritmo de previsão eficiente que acaba sendo quase ideal.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copiar
Ichiro TAJIKA, Eiji TAKIMOTO, Akira MARUOKA, "An On-Line Prediction Algorithm Combining Several Prediction Strategies in the Shared Bet Model" in IEICE TRANSACTIONS on Information,
vol. E82-D, no. 2, pp. 348-355, February 1999, doi: .
Abstract: One of the most important problems in machine learning is to predict a binary value by observing a sequence of outcomes, up to the present time step, generated from some unknown source. Vovk and Cesa-Bianchi et al. independently proposed an on-line prediction model where prediction algorithms are assumed to be given a collection of prediction strategies called experts and hence be allowed to use the predictions they make. In this model, no assumption is made about the way the sequence of bits to be predicted is generated, and the performance of the algorithm is measured by the difference between the number of mistakes it makes on the bit sequence and the number of mistakes made by the best expert on the same sequence. In this paper we extend the model by introducing a notion of investment. That is, both the prediction algorithm and the experts are required to make bets on their predictions at each time step, and the performance of the algorithm is now measured with respect to the total money lost, rather than the number of mistakes. We analyze our algorithms in the particular situation where all the experts share the same amount of bets at each time step. In this shared bet model, we give a prediction algorithm that is in some sense optimal but impractical, and we also give an efficient prediction algorithm that turns out to be nearly optimal.
URL: https://global.ieice.org/en_transactions/information/10.1587/e82-d_2_348/_p
Copiar
@ARTICLE{e82-d_2_348,
author={Ichiro TAJIKA, Eiji TAKIMOTO, Akira MARUOKA, },
journal={IEICE TRANSACTIONS on Information},
title={An On-Line Prediction Algorithm Combining Several Prediction Strategies in the Shared Bet Model},
year={1999},
volume={E82-D},
number={2},
pages={348-355},
abstract={One of the most important problems in machine learning is to predict a binary value by observing a sequence of outcomes, up to the present time step, generated from some unknown source. Vovk and Cesa-Bianchi et al. independently proposed an on-line prediction model where prediction algorithms are assumed to be given a collection of prediction strategies called experts and hence be allowed to use the predictions they make. In this model, no assumption is made about the way the sequence of bits to be predicted is generated, and the performance of the algorithm is measured by the difference between the number of mistakes it makes on the bit sequence and the number of mistakes made by the best expert on the same sequence. In this paper we extend the model by introducing a notion of investment. That is, both the prediction algorithm and the experts are required to make bets on their predictions at each time step, and the performance of the algorithm is now measured with respect to the total money lost, rather than the number of mistakes. We analyze our algorithms in the particular situation where all the experts share the same amount of bets at each time step. In this shared bet model, we give a prediction algorithm that is in some sense optimal but impractical, and we also give an efficient prediction algorithm that turns out to be nearly optimal.},
keywords={},
doi={},
ISSN={},
month={February},}
Copiar
TY - JOUR
TI - An On-Line Prediction Algorithm Combining Several Prediction Strategies in the Shared Bet Model
T2 - IEICE TRANSACTIONS on Information
SP - 348
EP - 355
AU - Ichiro TAJIKA
AU - Eiji TAKIMOTO
AU - Akira MARUOKA
PY - 1999
DO -
JO - IEICE TRANSACTIONS on Information
SN -
VL - E82-D
IS - 2
JA - IEICE TRANSACTIONS on Information
Y1 - February 1999
AB - One of the most important problems in machine learning is to predict a binary value by observing a sequence of outcomes, up to the present time step, generated from some unknown source. Vovk and Cesa-Bianchi et al. independently proposed an on-line prediction model where prediction algorithms are assumed to be given a collection of prediction strategies called experts and hence be allowed to use the predictions they make. In this model, no assumption is made about the way the sequence of bits to be predicted is generated, and the performance of the algorithm is measured by the difference between the number of mistakes it makes on the bit sequence and the number of mistakes made by the best expert on the same sequence. In this paper we extend the model by introducing a notion of investment. That is, both the prediction algorithm and the experts are required to make bets on their predictions at each time step, and the performance of the algorithm is now measured with respect to the total money lost, rather than the number of mistakes. We analyze our algorithms in the particular situation where all the experts share the same amount of bets at each time step. In this shared bet model, we give a prediction algorithm that is in some sense optimal but impractical, and we also give an efficient prediction algorithm that turns out to be nearly optimal.
ER -