### Abstract

Original language | English |
---|---|

Pages (from-to) | 1-21 |

Number of pages | 21 |

Journal | The Journal of Machine Learning Research |

Volume | 17 |

Issue number | 23 |

Publication status | Published - 1 Apr 2016 |

### Bibliographical note

© 2016 Dmitry Adamskiy, Wouter M. Koolen, Alexey Chernov and Vladimir Vovk### Keywords

- Online learning
- adaptive regret
- Fixed Share
- specialist experts

### Cite this

*The Journal of Machine Learning Research*,

*17*(23), 1-21.

}

*The Journal of Machine Learning Research*, vol. 17, no. 23, pp. 1-21.

**A closer look at adaptive regret.** / Adamskiy, Dmitry; Koolen, Wouter; Chernov, Alexey; Vovk, Vladimir.

Research output: Contribution to journal › Article › Research › peer-review

TY - JOUR

T1 - A closer look at adaptive regret

AU - Adamskiy, Dmitry

AU - Koolen, Wouter

AU - Chernov, Alexey

AU - Vovk, Vladimir

N1 - © 2016 Dmitry Adamskiy, Wouter M. Koolen, Alexey Chernov and Vladimir Vovk

PY - 2016/4/1

Y1 - 2016/4/1

N2 - For the prediction with expert advice setting, we consider methods to construct algorithms that have low adaptive regret. The adaptive regret of an algorithm on a time interval [t,T] is the loss of the algorithm minus the loss of the best expert over that interval. Adaptive regret measures how well the algorithm approximates the best expert locally, and so is different from, although closely related to, both the classical regret, measured over an initial time interval [1,t], and the tracking regret, where the algorithm is compared to a good sequence of experts over [1,t]. We investigate two existing intuitive methods for deriving algorithms with low adaptive regret, one based on specialist experts and the other based on restarts. Quite surprisingly, we show that both methods lead to the same algorithm, namely Fixed Share, which is known for its tracking regret. We provide a thorough analysis of the adaptive regret of Fixed Share. We obtain the exact worst-case adaptive regret for Fixed Share, from whichthe classical tracking bounds follow. We prove that Fixed Share is optimal for adaptive regret: the worst-case adaptive regret of any algorithm is at least that of an instance ofFixed Share.

AB - For the prediction with expert advice setting, we consider methods to construct algorithms that have low adaptive regret. The adaptive regret of an algorithm on a time interval [t,T] is the loss of the algorithm minus the loss of the best expert over that interval. Adaptive regret measures how well the algorithm approximates the best expert locally, and so is different from, although closely related to, both the classical regret, measured over an initial time interval [1,t], and the tracking regret, where the algorithm is compared to a good sequence of experts over [1,t]. We investigate two existing intuitive methods for deriving algorithms with low adaptive regret, one based on specialist experts and the other based on restarts. Quite surprisingly, we show that both methods lead to the same algorithm, namely Fixed Share, which is known for its tracking regret. We provide a thorough analysis of the adaptive regret of Fixed Share. We obtain the exact worst-case adaptive regret for Fixed Share, from whichthe classical tracking bounds follow. We prove that Fixed Share is optimal for adaptive regret: the worst-case adaptive regret of any algorithm is at least that of an instance ofFixed Share.

KW - Online learning

KW - adaptive regret

KW - Fixed Share

KW - specialist experts

M3 - Article

VL - 17

SP - 1

EP - 21

JO - The Journal of Machine Learning Research

JF - The Journal of Machine Learning Research

SN - 1532-4435

IS - 23

ER -