Structure Adaptive Algorithms for Stochastic Bandits

Authors: Rémy Degenne, Han Shao, Wouter Koolen

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We perform four series of experiments. In all cases, rewards are Gaussian with variance 1. The structures are: ... We mainly compare the following four algorithms, SPk, our saddle point algorithm based on a k-learner. ... Figure 1 reports the mean regret of these algorithms over 200 repetitions.
Researcher Affiliation Academia INRIA DIENS PSL Research University, Paris, France 2Toyota Technological Institute at Chicago 3Centrum Wiskunde & Informatica.
Pseudocode Yes Algorithm 1 SPk Learner
Open Source Code No The paper does not contain any explicit statement about releasing the source code for the described methodology or a link to a code repository.
Open Datasets No The paper describes setting up experiments with simulated data ('rewards are Gaussian with variance 1') and specific structural parameters, but does not refer to any pre-existing publicly available datasets, nor does it provide access information for any self-generated dataset.
Dataset Splits No The paper mentions running experiments and reporting 'mean regret over 200 repetitions' but does not specify any train/validation/test dataset splits or cross-validation methods.
Hardware Specification No The paper does not provide any specific hardware details such as GPU models, CPU types, or memory specifications used for running the experiments.
Software Dependencies No The paper mentions general algorithms like Ada Hedge and Follow-the-Leader, but does not provide specific software names with version numbers for reproducibility.
Experiment Setup No The paper states 'Implementation details are in Appendix G' in Section 3, but Appendix G is not provided in the given text. Thus, no specific hyperparameters or detailed training configurations are present in the main body.