Multiplier Bootstrap-based Exploration

Authors: Runzhe Wan, Haoyu Wei, Branislav Kveton, Rui Song

ICML 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental With extensive simulation and real-data experiments, we show the generality and adaptivity of MBE.
Researcher Affiliation Collaboration *Equal contribution 1Amazon 2Department of Economics, University of California San Diego 3Department of Statistics, North Carolina State University. Correspondence to: Runzhe Wan <runzhe.wan@gmail.com>, Rui Song <songray@gmail.com>.
Pseudocode Yes Algorithm 1: General Template for MBE
Open Source Code No The paper does not explicitly state that source code for the methodology is being released or provide a link to a code repository.
Open Datasets Yes We use the Yelp rating dataset (Zong et al., 2016) to recommend and rank K restaurants, use the Adult dataset (Dua & Graff, 2017) to send advertisements to K/2 men and K/2 women (a combinatorial semi-bandit problem with continuous rewards), and use the Movie Lens dataset (Harper & Konstan, 2015) to display K movies.
Dataset Splits No The paper mentions splitting datasets into training and testing sets but does not provide specific split percentages, absolute sample counts, or refer to predefined splits with citations for reproducibility.
Hardware Specification No The paper does not provide specific hardware details such as CPU/GPU models, memory, or cloud instance types used for running experiments.
Software Dependencies No The paper does not explicitly list software dependencies with version numbers.
Experiment Setup Yes In all experiments below, the weights of MBE are sampled from N(1, σ2 ω) 1. We fix λ = 0.5 and run MBE with three different values of σ2 ω: 0.5, 1 and 1.5. We also compare with the naive adaption of multiplier bootstrap (i.e., no pseudorewards; denoted as Naive MB). We run Algorithm 2 with B = 50 replicates.