Non-Asymptotic Pure Exploration by Solving Games

Authors: Rémy Degenne, Wouter M. Koolen, Pierre Ménard

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

Reproducibility Variable Result LLM Response
Research Type Experimental We validate our approach empirically in benchmark experiments at practical δ, and find that our algorithms are either competitive with Track-and-Stop (dense w ) or dominate it (sparse w ).
Researcher Affiliation Academia Rémy Degenne Centrum Wiskunde & Informatica Science Park 123, 1098 XG Amsterdam remy.degenne@cwi.nl
Pseudocode Yes Algorithm 1 Pure exploration meta-algorithm.
Open Source Code No The paper does not provide any statement or link indicating the availability of open-source code for the described methodology.
Open Datasets No The paper describes experiments on 'Bernoulli bandit model' and 'Gaussian bandit model' with specified parameters, indicating a simulated environment rather than the use of a pre-existing publicly available dataset that would require a link or citation for access.
Dataset Splits No The paper operates within a multi-armed bandit framework where data is sampled sequentially, and therefore, does not discuss traditional training, validation, and test dataset splits as found in static dataset-based experiments.
Hardware Specification No The paper states 'The experiments were carried out on the Dutch national e-infrastructure with the support of SURF Cooperative,' but it does not provide specific hardware details such as GPU/CPU models, memory, or processor types.
Software Dependencies No The paper does not specify any software dependencies with version numbers required to replicate the experiments.
Experiment Setup Yes We use stylised stopping threshold β(δ, t) = ln 1+ln t / δ and exploration bonus f(t) = ln t. Both are unlicensed by theory yet conservative in practise (the error frequency is way below δ).