Zobrazeno 1 - 10
of 26
pro vyhledávání: '"Yevgeny Seldin"'
Autor:
Saeed Masoudian, Yevgeny Seldin
Publikováno v:
University of Copenhagen
Masoudian, S & Seldin, Y 2021, Improved Analysis of the Tsallis-INF Algorithm in Stochastically Constrained Adversarial Bandits and Stochastic Bandits with Adversarial Corruptions . in Conference on Learning Theory (COLT 2021) . PMLR, Proceedings of Machine Learning Research, vol. 134, pp. 3330-3350, 34th Annual Conference on Learning Theory (COLT 2021), Boulder, Colorado, United States, 15/08/2021 .
Masoudian, S & Seldin, Y 2021, Improved Analysis of the Tsallis-INF Algorithm in Stochastically Constrained Adversarial Bandits and Stochastic Bandits with Adversarial Corruptions . in Conference on Learning Theory (COLT 2021) . PMLR, Proceedings of Machine Learning Research, vol. 134, pp. 3330-3350, 34th Annual Conference on Learning Theory (COLT 2021), Boulder, Colorado, United States, 15/08/2021 .
We derive improved regret bounds for the Tsallis-INF algorithm of Zimmert and Seldin (2021). We show that in adversarial regimes with a $(\Delta,C,T)$ self-bounding constraint the algorithm achieves $\mathcal{O}\left(\left(\sum_{i\neq i^*} \frac{1}{\
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::861bc03056df2f1992de647822b87554
http://arxiv.org/abs/2103.12487
http://arxiv.org/abs/2103.12487
Publikováno v:
University of Copenhagen
Rouyer, C, Seldin, Y & Cesa-Bianchi, N 2021, An algorithm for stochastic and adversarial bandits with switching costs . in Proceedings of the 38th International Conference on Machine Learning (ICML) . PMLR, Proceedings of Machine Learning Research, vol. 139, pp. 9127-9135, 38th International Conference on Machine Learning (ICML), Virtual, 18/07/2021 . < https://proceedings.mlr.press/v139/ >
Rouyer, C, Seldin, Y & Cesa-Bianchi, N 2021, An algorithm for stochastic and adversarial bandits with switching costs . in Proceedings of the 38th International Conference on Machine Learning (ICML) . PMLR, Proceedings of Machine Learning Research, vol. 139, pp. 9127-9135, 38th International Conference on Machine Learning (ICML), Virtual, 18/07/2021 . < https://proceedings.mlr.press/v139/ >
We propose an algorithm for stochastic and adversarial multiarmed bandits with switching costs, where the algorithm pays a price $\lambda$ every time it switches the arm being played. Our algorithm is based on adaptation of the Tsallis-INF algorithm
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::37894ff5faf092774d9cf65d0ee4f732
http://arxiv.org/abs/2102.09864
http://arxiv.org/abs/2102.09864
Publikováno v:
Wu, Y-S, Masegosa, A, Lorenzen, S S, Igel, C & Seldin, Y 2021, Chebyshev-Cantelli PAC-Bayes-Bennett Inequality for the Weighted Majority Vote . in Advances in Neural Information Processing Systems (NeurIPS 2021) . vol. 34, Thirty-fifth Conference on Neural Information Processing Systems-NeurIPS 2021, 06/12/2021 . < https://proceedings.neurips.cc/paper/2021/hash/69386f6bb1dfed68692a24c8686939b9-Abstract.html >
University of Copenhagen
University of Copenhagen
We present a new second-order oracle bound for the expected risk of a weighted majority vote. The bound is based on a novel parametric form of the Chebyshev- Cantelli inequality (a.k.a. one-sided Chebyshev's), which is amenable to efficient minimizat
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::9387c162dd73fec6af0a01654c0c747a
https://vbn.aau.dk/da/publications/2b14b130-cd23-4fd1-b655-9418de9e9f57
https://vbn.aau.dk/da/publications/2b14b130-cd23-4fd1-b655-9418de9e9f57
Publikováno v:
University of Copenhagen
Masegosa, A R, Lorenzen, S S, Igel, C & Seldin, Y 2020, Second Order PAC-Bayesian Bounds for the Weighted Majority Vote . in Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual . NeurIPS Proceedings, Advances in Neural Information Processing Systems, vol. 33, 34th Conference on Neural Information Processing System (NeurIPS 2020), Virtuak, 06/12/2020 . < https://proceedings.neurips.cc/paper/2020/file/386854131f58a556343e056f03626e00-Paper.pdf >
Masegosa, A R, Lorenzen, S S, Igel, C & Seldin, Y 2020, Second Order PAC-Bayesian Bounds for the Weighted Majority Vote . in Advances in Neural Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, December 6-12, 2020, virtual . NeurIPS Proceedings, Advances in Neural Information Processing Systems, vol. 33, 34th Conference on Neural Information Processing System (NeurIPS 2020), Virtuak, 06/12/2020 . < https://proceedings.neurips.cc/paper/2020/file/386854131f58a556343e056f03626e00-Paper.pdf >
We present a novel analysis of the expected risk of weighted majority vote in multiclass classification. The analysis takes correlation of predictions by ensemble members into account and provides a bound that is amenable to efficient minimization, w
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::f53996f234fd39e982f600d10d02409d
http://arxiv.org/abs/2007.13532
http://arxiv.org/abs/2007.13532
Autor:
Chloé Rouyer, Yevgeny Seldin
Publikováno v:
Rouyer, C & Seldin, Y 2020, Tsallis-INF for decoupled exploration and exploitation in multi-armed bandits . in Proceedings of Thirty Third Conference on Learning Theory(COLT) . PMLR, Proceedings of Machine Learning Research, vol. 125, pp. 3227-3249 . < https://proceedings.mlr.press/v125/ >
University of Copenhagen
University of Copenhagen
We consider a variation of the multi-armed bandit problem, introduced by Avner et al. (2012), in which the forecaster is allowed to choose one arm to explore and one arm to exploit at every round. The loss of the exploited arm is blindly suffered by
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=dedup_wf_001::7f476e03d56edb4df2e0287652b3c205
https://curis.ku.dk/ws/files/323980162/tsallis_inf_for_decoupled.pdf
https://curis.ku.dk/ws/files/323980162/tsallis_inf_for_decoupled.pdf
Autor:
Julian Ulf Zimmert, Yevgeny Seldin
Publikováno v:
University of Copenhagen
Zimmert, J U & Seldin, Y 2020, An optimal algorithm for adversarial bandits with arbitrary delays . in Proceedings of the 23rdInternational Conference on Artificial Intelligence and Statistics (AISTATS) 2020 . PMLR, Proceedings of Machine Learning Research, vol. 108 . < https://proceedings.mlr.press/v108/ >
Zimmert, J U & Seldin, Y 2020, An optimal algorithm for adversarial bandits with arbitrary delays . in Proceedings of the 23rdInternational Conference on Artificial Intelligence and Statistics (AISTATS) 2020 . PMLR, Proceedings of Machine Learning Research, vol. 108 . < https://proceedings.mlr.press/v108/ >
We propose a new algorithm for adversarial multi-armed bandits with unrestricted delays. The algorithm is based on a novel hybrid regularizer applied in the Follow the Regularized Leader (FTRL) framework. It achieves $\mathcal{O}(\sqrt{kn}+\sqrt{D\lo
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::500139715e96edfb2d08daf20e9e8e43
http://arxiv.org/abs/1910.06054
http://arxiv.org/abs/1910.06054
Publikováno v:
University of Copenhagen
We investigate multiarmed bandits with delayed feedback, where the delays need neither be identical nor bounded. We first prove that "delayed" Exp3 achieves the $O(\sqrt{(KT + D)\ln K} )$ regret bound conjectured by Cesa-Bianchi et al. [2019] in the
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::a4cf81a4f195f66089745ae198b373d8
http://arxiv.org/abs/1906.00670
http://arxiv.org/abs/1906.00670
Autor:
Julian Ulf Zimmert, Yevgeny Seldin
Publikováno v:
Zimmert, J U & Seldin, Y 2021, ' Tsallis-INF: An optimal algorithm for stochastic and adversarial bandits ', Journal of Machine Learning Research, vol. 22, 28 .
University of Copenhagen
University of Copenhagen
We derive an algorithm that achieves the optimal (within constants) pseudo-regret in both adversarial and stochastic multi-armed bandits without prior knowledge of the regime and time horizon. The algorithm is based on online mirror descent (OMD) wit
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::82186c275c190d04b0b20e118faedec3
http://arxiv.org/abs/1807.07623
http://arxiv.org/abs/1807.07623
Existing guarantees in terms of rigorous upper bounds on the generalization error for the original random forest algorithm, one of the most frequently used machine learning methods, are unsatisfying. We discuss and evaluate various PAC-Bayesian appro
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=doi_dedup___::cb114bcee500a633a2203b2c69cd1b0c
Autor:
Yevgeny Seldin, Gábor Lugosi
Publikováno v:
University of Copenhagen
We present a new strategy for gap estimation in randomized algorithms for multiarmed bandits and combine it with the EXP3++ algorithm of Seldin and Slivkins (2014). In the stochastic regime the strategy reduces dependence of regret on a time horizon
Externí odkaz:
https://explore.openaire.eu/search/publication?articleId=arXiv_dedup_::5848ee038cbe902b6edd57fdea6f9b34
http://arxiv.org/abs/1702.06103
http://arxiv.org/abs/1702.06103