Online Resource Allocation with Non-Stationary Customers

Authors: Xiaoyue Zhang, Hanzhang Qin, Mabel Chou

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

Reproducibility Variable Result LLM Response
Research Type Experimental Finally, we conduct extensive numerical experiments to show our approach generates near-optimal revenues for all different customer scenarios.
Researcher Affiliation Academia Institute of Operations Research and Analytics, National University of Singapore, Singapore 117602.
Pseudocode Yes We propose the ULwE algorithm, displayed in Algorithm 1, which is applicable in both stochastic and adversarial environments.
Open Source Code No The paper does not include any statement about releasing source code for the methodology or a link to a code repository.
Open Datasets No The experiments use simulated data rather than a publicly available dataset with a provided link or citation. The paper states: "Initially, based on the true rate µ, we generate data for 500 customers to estimate the initial λl values".
Dataset Splits No The paper mentions generating data for experiments but does not specify any training, validation, or test dataset splits. It describes how initial values are estimated and how `λls` are updated, but not data partitioning for model training/evaluation.
Hardware Specification No The paper describes the experimental setup but does not provide specific details about the hardware used (e.g., GPU/CPU models, memory specifications).
Software Dependencies No The paper does not specify any software dependencies with version numbers used for implementing or running the experiments.
Experiment Setup Yes The experiments involve two customer types (L = 2) and two resources (n = 2), with customer purchase probabilities modeled through a logistic function...Customer Type A exhibits a high purchase probability of 0.9, while Customer Type B maintains a consistent probability of 0.5...The total capacity of both resources matches the time horizon T, with revenues set at 1 and 1.5 units for Resource 1 and 2, respectively. Our analysis primarily assesses the ULwE algorithm s performance, particularly its regret in both i.i.d. and adversarial customer arrival scenarios, highlighting its adaptability across various operational contexts. Experiment s regret values represent the average outcomes of 100 independent experiments, ensuring robustness and reliability of the results.