Bootstrapping Upper Confidence Bound

Authors: Botao Hao, Yasin Abbasi Yadkori, Zheng Wen, Guang Cheng

NeurIPS 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Numerical results demonstrate significant regret reductions by our method, in comparison with several baselines in a range of multi-armed and linear bandit problems.
Researcher Affiliation Collaboration Botao Hao Purdue University haobotao000@gmail.com Yasin Abbasi-Yadkori Vin AI yasin.abbasi@gmail.com Zheng Wen Deepmind zhengwen@google.com Guang Cheng Purdue University chengg@purdue.edu
Pseudocode Yes Algorithm 1 Bootstrapped UCB
Open Source Code No The paper does not provide any links to or explicit statements about the release of its source code.
Open Datasets No The paper describes generating data from distributions (e.g., truncated-normal, standard Gaussian, Bernoulli, Beta) rather than using specific, publicly available datasets with citations or links.
Dataset Splits No The paper does not provide specific training/test/validation dataset splits, sample counts, or cross-validation methods.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running its experiments.
Software Dependencies No The paper mentions the
Experiment Setup Yes To be fair, we choose the confidence level α = 1/(1 + t) for both UCB1 and bootstrapped UCB, and δ = 0.1 in (2.5). All algorithms above require knowledge of an upper bound on the noise standard deviation. The number of bootstrap repetitions is B = 200, and the number of arms is K = 5.