Learning the Optimal Recommendation from Explorative Users

Authors: Fan Yao, Chuanhao Li, Denis Nekipelov, Hongning Wang, Haifeng Xu9457-9465

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we empirically study BAIR to support our theoretical analysis. We use simulations on synthetic datasets in comparison with several baseline algorithms. The results are reported in Table 2. Based on the comparison results for BAIR and the baselines, we have the following observations.
Researcher Affiliation Academia 1Department of Computer Science, University of Virginia, USA 2Department of Economics, University of Virginia, USA
Pseudocode Yes Algorithm 1: Phase-1 Sweeping and Algorithm 2: Phase-2 Elimination
Open Source Code No The paper does not provide a specific link to source code or an explicit statement about its release.
Open Datasets No We use simulations on synthetic datasets in comparison with several baseline algorithms. The paper does not provide access information (link, DOI, specific citation with authors/year) for these synthetic datasets.
Dataset Splits No The paper uses synthetic datasets but does not explicitly provide training/test/validation splits (percentages, counts, or specific predefined split citations).
Hardware Specification No The paper mentions running simulations but does not provide any specific details about the hardware (e.g., CPU, GPU models) used for these experiments.
Software Dependencies No The paper does not provide specific software dependencies, such as library names with version numbers, used to replicate the experiment.
Experiment Setup Yes For different configurations of (δ, K, Δ1) for BAI, we generate 1000 independent problem instances (μi)Ki=1 by sampling each μi ∼ N(0, 1) and then reset μ to meet the given value of Δ1. Observing that our conclusion does not vary much under different Δ1, we present the result for Δ1 = 0.5 in this section... The parameters in the user model are set to α = 1, ρt = 1 + n(t)/t ∈ [1, 2], i.e., ρ0 = 1, ρ1 = 2... We run BAIR with N1 = 2K/δ and compare its performance with UNI, EXP3 and T&S...