Graph-Triggered Rising Bandits

Authors: Gianmarco Genalti, Marco Mussi, Nicola Gatti, Marcello Restelli, Matteo Castiglioni, Alberto Maria Metelli

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this appendix, we provide an experimental campaign to validate the proposed algorithmic solutions from an empirical perspective. We start with the deterministic setting: in Appendix D.1 we evaluate DR-BG-UB in 15 GTRB instances, varying both the functions and the adjacency matrices; in Appendix D.2 we evaluate DR-G-UB in 3 GTRB instances, but varying the sub-matrix used in the Algorithm 2 routine. Finally, we evaluate R-UCB in 10 stochastic GTRB instances, varying both the functions and the adjacency matrices, and comparing its performances to a baseline from the literature, Sliding Window UCB (Garivier & Moulines, 2011).
Researcher Affiliation Academia 1Politecnico di Milano, Milan, Italy. Correspondence to: G. Genalti <gianmarco.genalti@polimi.it>.
Pseudocode Yes Algorithm 1: DR-BG-UB. Algorithm 2: DR-G-UB. Algorithm 3: R-UCB.
Open Source Code No The paper does not provide any concrete access to source code for the methodology described, such as a specific repository link or an explicit code release statement.
Open Datasets No The paper conducts experiments on synthetic environments using 'sets of functions' (Figure 2, Figure 4) rather than referencing established public datasets with access information or formal citations.
Dataset Splits No The paper describes synthetic experimental settings but does not provide specific dataset split information (percentages, sample counts, or citations to predefined splits) for training, validation, and testing.
Hardware Specification No The paper does not provide specific hardware details (exact GPU/CPU models, processor types, or memory amounts) used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details, such as library or solver names with version numbers, needed to replicate the experiments.
Experiment Setup Yes The hyper-parameters of SW-UCB have been set according to the original paper (Garivier & Moulines, 2011) and then optimized to get the smaller regret upper bound. Instead, the hyper-parameters of R-UCB have been fixed for all experiments, using ϵ = 0.1 and α = 3.