Collaborative Pure Exploration in Kernel Bandit

Authors: Yihan Du, Wei Chen, Yuko Kuroki, Longbo Huang

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we provide the experimental results. Here we set V = 5, n = 6, δ = 0.005, H = Rd, d {4, 8, 20}, θ = [0.1, 0.1 + Δmin, . . . , 0.1 + (d 1)Δmin] , Δmin [0.1, 0.8] and rank(KZ) [1, V ]. We run 50 independent simulations and plot the average sample complexity with 95% confidence intervals (see Appendix A for a complete setup description and more results).
Researcher Affiliation Collaboration Yihan Du Institute for Interdisciplinary Information Sciences Tsinghua University Beijing, China duyh18@mails.tsinghua.edu.cn Wei Chen Microsoft Research Beijing, China weic@microsoft.com Yuko Kuroki The University of Tokyo & RIKEN AIP, Tokyo, Japan CENTAI Institute, Turin, Italy yuko.kuroki@centai.eu Longbo Huang Institute for Interdisciplinary Information Sciences Tsinghua University Beijing, China longbohuang@tsinghua.edu.cn
Pseudocode Yes Algorithm 1 Collaborative Multi-agent Algorithm Co Kernel FC: for Agent v [V ] [...] Algorithm 2 Collaborative Multi-agent Algorithm Co Kernel FB: for Agent v [V ]
Open Source Code No The paper does not contain any explicit statement about providing open-source code or a link to a code repository.
Open Datasets No The paper describes a synthetic experimental setup where parameters like V, n, δ, H, d, θ, Δmin, and rank(KZ) are set or tuned to generate instances for simulations, rather than using a pre-existing publicly available dataset. There is no mention of an external dataset being used that would require an access link or citation.
Dataset Splits No The paper describes generating different instances for simulations and running independent simulations (50 or 100 runs) to plot average metrics. This is not the same as providing specific train/validation/test dataset splits for reproducibility, as the data is generated within the simulation rather than partitioned from a fixed dataset.
Hardware Specification Yes Our experiments are run on Intel Xeon E5-2660 v3 CPU at 2.60GHz.
Software Dependencies No The paper does not provide specific ancillary software details with version numbers (e.g., Python, PyTorch, specific libraries or solvers). It only describes the conceptual algorithms and experimental setup parameters.
Experiment Setup Yes Here we set V = 5, n = 6, δ = 0.005, H = Rd, d {4, 8, 20}, θ = [0.1, 0.1 + Δmin, . . . , 0.1 + (d 1)Δmin] , Δmin [0.1, 0.8] and rank(KZ) [1, V ]. We run 50 independent simulations and plot the average sample complexity with 95% confidence intervals (see Appendix A for a complete setup description and more results). [...] For the rank(KZ) = 1 case, we set d = 4. For any v [V ], {φ( x)} x Xv is the set of all 4 2 vectors in R4, where each vector has two entries 0 and two entries 1.