Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in [1].

VCG Mechanism Design with Unknown Agent Values under Stochastic Bandit Feedback

Authors: Kirthevasan Kandasamy, Joseph E Gonzalez, Michael I Jordan, Ion Stoica

JMLR 2023 | Venue PDF | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We present some simulation results in a single-parameter single-item environment... We have shown the pseudo-regrets for the welfare, the seller, and some of the agents in Figure 1 for all possible choices of the ζ and λ hyperparameters.
Researcher Affiliation Academia Kirthevasan Kandasamy EMAIL University of Wisconsin Madison, WI 53706, USA. Joseph E Gonzalez EMAIL Michael I Jordan EMAIL Ion Stoica EMAIL University of California, Berkeley, CA 94723, USA
Pseudocode Yes We now describe our algorithm for this setting, called VCG-Learn, which is outlined in Algorithm 1. ... Algorithm 1 VCG-Learn
Open Source Code No The paper does not provide concrete access to source code for the methodology described.
Open Datasets No We present some simulation results in a single-parameter single-item environment. Here, ten agents are competing for a single item and all of them are participating by rewards. When an agent receives the item, her value is drawn stochastically from a N(µ, 0.5) distribution where µ is chosen uniformly on a grid in the interval (0.2, 0.9).
Dataset Splits No The paper uses a simulated environment where data is generated stochastically; therefore, traditional training/test/validation dataset splits are not applicable or provided.
Hardware Specification No The paper does not provide specific hardware details used for running its experiments.
Software Dependencies No The paper does not provide specific ancillary software details, such as library names with version numbers.
Experiment Setup Yes We present some simulation results in a single-parameter single-item environment. Here, ten agents are competing for a single item and all of them are participating by rewards. When an agent receives the item, her value is drawn stochastically from a N(µ, 0.5) distribution where µ is chosen uniformly on a grid in the interval (0.2, 0.9). Agent 1 has a value of 0.9 for receiving the item (and will be the agent who receives the item if values are known) and agent 10 has a value of 0.2. If an agent does not receive the item, their value is non-stochastically zero. ... The game is repeated for 3000 rounds. ... The figures were obtained by averaging over 50 independent runs and the shaded regions represent two standard errors.