Decentralized Randomly Distributed Multi-agent Multi-armed Bandit with Heterogeneous Rewards
Authors: Mengfan Xu, Diego Klabjan
NeurIPS 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | In this section, we present a numerical study of the proposed algorithm. Specifically, we first demonstrate the regret performance of Algorithms 2 and 3, in comparison with existing benchmark methods from the literature, in a setting with time-invariant graphs. Moreover, we conduct a numerical experiment with respect to time-varying graphs, comparing the proposed algorithm with the most recent work [Zhu and Liu, 2023]. |
| Researcher Affiliation | Academia | 1Department of Industrial Engineering and Management Sciences, Northwestern University Mengfan Xu2023@u.northwestern.edu, d-klabjan@northwestern.edu |
| Pseudocode | Yes | Algorithm 1: Dr Fed-UCB: Burn-in period; Algorithm 2: Dr Fed-UCB: Learning period; Algorithm 3: Generate a uniformly distributed connected graph |
| Open Source Code | No | The paper acknowledges that code for benchmark algorithms was shared by other authors: 'Additionally, we are much obliged to the authors of the papers [Chawla et al., 2020, Zhu et al., 2021b]... for promptly sharing the code of their algorithms, which has helped us to run the benchmarks presented in this work.' However, there is no statement or link indicating that the authors' own code for the proposed Dr Fed-UCB method is open-source or publicly available. |
| Open Datasets | No | The paper states: 'First we generate different numbers of arms and clients, denoted as K and M, respectively. Specifically, we generate rewards using the Bernoulli distribution in the sub-Gaussian distribution family... In terms of graph generation, we generate E-R models...'. This indicates data was generated by the authors, not from a publicly available dataset with concrete access information. |
| Dataset Splits | No | The paper mentions a 'burn-in period' in its algorithm but does not describe conventional training, validation, or test dataset splits (e.g., 80/10/10 split or k-fold cross-validation) for data partitioning. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used for running the experiments, such as GPU or CPU models, memory, or cloud computing specifications. |
| Software Dependencies | No | The paper does not list specific software dependencies with version numbers (e.g., Python version, library versions) that would be needed to reproduce the experiments. |
| Experiment Setup | No | Appendix C, 'Details on numerical experiments in Section 4', describes parameters related to data and graph generation (e.g., K, M, h, c for problem settings) and notes the comparison with benchmark algorithms. However, it does not specify hyperparameters of the proposed Dr Fed-UCB algorithm itself, such as learning rates, batch sizes, number of epochs, or optimizer settings, which are common details for experimental setup reproducibility. |