Maximizing and Satisficing in Multi-armed Bandits with Graph Information

Authors: Parth Thaker, Mohit Malu, Nikhil Rao, Gautam Dasarathy

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

Reproducibility Variable Result LLM Response
Research Type Experimental We show additional experiments with different graph parameters for Stochastic block model and Barabási-Albert graphs and different cluster sizes as well as real data in Appendix K. The full code used for conducting experiments can be found at the following Github repository. For all our experiments, we use Intel R Core TM i7-10875H CPU @ 2.30GHz 16 with 32 GB memory. We set δ = 1e 3, ρ = 2.0, σ = 2.0. We evaluate GRUB with different sampling strategies from section J and compare its performance to standard UCB algorithm on both synthetic and real datasets.
Researcher Affiliation Collaboration Parth K. Thaker Arizona State University pkthaker@asu.edu Mohit Malu Arizona State University mmalu@asu.edu Nikhil Rao Microsoft nikhilrao86@gmail.com Gautam Dasarathy Arizona State University gautamd@asu.edu
Pseudocode Yes The pseudocode for GRUB can be found in Appendix E. The pseudocode for the ζ-GRUB can be found in Appendix G.
Open Source Code Yes The full code used for conducting experiments can be found at the following Github repository.
Open Datasets No The paper states it uses 'Synthetic Data' generated from models like 'Stochastic Block Model' and 'Barabási Albert graph', and 'real data in Appendix K'. While the ethics checklist mentions 'Data is public', the main text does not provide specific citations, links, or names of well-known public datasets for which access information is directly provided.
Dataset Splits No The paper does not provide specific dataset split information (e.g., percentages or counts for training, validation, or test sets). It mentions 'training details' in the checklist and parameters like δ, ρ, σ, but not how the data was partitioned into splits for reproduction.
Hardware Specification Yes For all our experiments, we use Intel R Core TM i7-10875H CPU @ 2.30GHz 16 with 32 GB memory.
Software Dependencies No The paper does not provide specific software dependencies with version numbers (e.g., library or solver names with version numbers) needed to replicate the experiment.
Experiment Setup Yes For all our experiments, we use Intel R Core TM i7-10875H CPU @ 2.30GHz 16 with 32 GB memory. We set δ = 1e 3, ρ = 2.0, σ = 2.0.