Adaptive Experimentation When You Can't Experiment
Authors: Yao Zhao, Kwang-Sung Jun, Tanner Fiez, Lalit Jain
NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Finally, experiments are conducted that demonstrate the efficacy of our approach. Though the goal of this work is primarily theoretical, we empirically show the efficacy of our method over existing solutions (Appendix E). |
| Researcher Affiliation | Collaboration | Yao Zhao University of Arizona yaoz@arizona.edu Kwang-Sung Jun University of Arizona kjun@cs.arizona.edu Tanner Fiez Amazon fieztann@amazon.com Lalit Jain University of Washington lalitj@uw.edu |
| Pseudocode | Yes | Algorithm 1 CPEG:Confounded pure exploration with Γ ... Algorithm 3 CPEUG: Confounded pure exploration with with unknown Γ ... Algorithm 2 θ estimator ... Algorithm 4 Γ estimator ... Algorithm 10 Learning λmin(Γ) |
| Open Source Code | No | While the NeurIPS checklist indicates that code is attached for simulation, the main body of the paper does not contain an explicit statement or a link to a code repository. |
| Open Datasets | No | The paper uses simulated data instances for its experiments, rather than a publicly available dataset. It defines the parameters for generating these simulated instances (e.g., 'd = 6, θ = 1 0.95 0 0.45 0.95 0.99 , and σ2 u = 0.35'). |
| Dataset Splits | No | The paper describes simulated experiments with specific parameter settings, but it does not specify training, validation, or test dataset splits in the context of machine learning model development. |
| Hardware Specification | No | The paper states it is 'mainly a theoretical paper' with 'some simulations in appendix' and does not provide any specific details about the hardware used to run these simulations. |
| Software Dependencies | No | The paper does not explicitly list any software dependencies with specific version numbers for its experiments. |
| Experiment Setup | Yes | Section B 'Illustrative Example' and Section E 'Experiments' detail the setup of their simulation instances. For example, 'Specifically, d = 6, θ = 1 0.95 0 0.45 0.95 0.99 , and σ2 u = 0.35.' is provided as part of the instance definition for the experiments. Section E.1 also discusses the 'Comparison Algorithms' and their round structure. |