Contextual Bandits with Knapsacks for a Conversion Model

Authors: Zhen LI, Gilles Stoltz

NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental A simulation study on partially simulated but realistic data may be found in Appendix F. The underlying dataset is the standard default of credit card clients dataset of UCI [2016], initially provided by Yeh and Lien [2009].
Researcher Affiliation Collaboration Zhen Li BNP Paribas, 16 boulevard des Italiens, 75009 Paris, France zhen.li@bnpparibas.com Gilles Stoltz Université Paris-Saclay, CNRS, Laboratoire de mathématiques d Orsay, 91405, Orsay, France gilles.stoltz@universite-paris-saclay.fr HEC Paris, 1 rue de la Libération, 78350 Jouy-en-Josas, France stoltz@hec.fr
Pseudocode Yes BOX B: LOGISTIC-UCB1 FOR DIRECT SOLUTIONS TO OPT PROBLEMS
Open Source Code Yes (a) Did you include the code, data, and instructions needed to reproduce the main experimental results (either in the supplemental material or as a URL)? [Yes]
Open Datasets Yes The underlying dataset is the standard default of credit card clients dataset of UCI [2016], initially provided by Yeh and Lien [2009]. (It may be used under a Creative Commons Attribution 4.0 International [CC BY 4.0] license.)
Dataset Splits No The paper mentions a simulation study using the UCI dataset in Appendix F, and the authors state in the ethics checklist (3.b) that they specified training details including data splits. However, Appendix F itself does not explicitly provide percentages, sample counts, or specific methods for train/validation/test splits.
Hardware Specification Yes We ran our simulations on a machine with an Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz processor, and with 16GB of RAM.
Software Dependencies No The paper does not provide specific version numbers for software dependencies (e.g., programming languages, libraries, frameworks, or solvers used in the implementation).
Experiment Setup Yes Parameters: regularization parameter λ > 0; conservative-budget parameter BT; upperconfidence bonuses εs(a, x) > 0, for s ≥ 1 and (a, x) ∈ A {anull} × X. ... We set a confidence level 1 − δ ∈ (0, 1) and use parameters λ = m ln(1 + T/m), 2T ln(4d/δ) , and εt(a, x) stated in (9) of the supplementary material.