An Adaptive Approach for Infinitely Many-armed Bandits under Generalized Rotting Constraints

Authors: Jung-hun Kim, Milan Vojnovic, Se-Young Yun

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

Reproducibility Variable Result LLM Response
Research Type Experimental Lastly, we demonstrate the performance of our algorithm using numerical experiments.
Researcher Affiliation Academia Jung-hun Kim Seoul National University Seoul, South Korea junghunkim@snu.ac.kr Milan Vojnovic London School of Economics London, United Kingdom m.vojnovic@lse.ac.uk Se-Young Yun KAIST AI Seoul, South Korea yunseyoung@kaist.ac.kr
Pseudocode Yes Algorithm 1 UCB-Threshold with Adaptive Sliding Window
Open Source Code Yes The source code is available at https://github.com/junghunkim7786/An-Adaptive-Approach-for-Infinitely-Many-armed-Bandits-under-Generalized-Rotting-Constraints
Open Datasets No We use randomly generated datasets under a uniform distribution for initial mean rewards (β = 1). The paper does not provide a link, DOI, or formal citation for this dataset as it is synthetically generated.
Dataset Splits No The paper mentions using 'randomly generated datasets' but does not specify any training, validation, or test split percentages or methodology.
Hardware Specification No The paper does not provide any specific details about the hardware used for running the experiments (e.g., GPU models, CPU types, memory).
Software Dependencies No The paper does not specify any software dependencies with version numbers (e.g., Python, PyTorch, or specific libraries).
Experiment Setup Yes We use randomly generated datasets under a uniform distribution for initial mean rewards (β = 1). For comparison with UCB-TP, recall our discussion in Remark 3.2. We set the rotting rates such that ρt = 1/(t log(T)) for all t, for which ρ = ρ1 = 1/ log(T) = o(1), VT = O(1), and ST = T.