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].
Misspecified Gaussian Process Bandit Optimization
Authors: Ilija Bogunovic, Andreas Krause
NeurIPS 2021 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Theoretical | We design efficient and practical algorithms whose performance degrades minimally in the presence of model misspecification. Specifically, we present two algorithms based on Gaussian process (GP) methods: an optimistic EC-GP-UCB algorithm that requires knowing the misspecification error, and Phased GP Uncertainty Sampling, an elimination-type algorithm that can adapt to unknown model misspecification. We provide upper bounds on their cumulative regret in terms of , the time horizon, and the underlying kernel, and we show that our algorithm achieves optimal dependence on with no prior knowledge of misspecification. |
| Researcher Affiliation | Academia | Ilija Bogunovic Andreas Krause (Authors listed on page 1) This project has received funding from the European Research Council (ERC) under the European Unions Horizon 2020 research and innovation programme grant agreement No 815943 and ETH Zรผrich Postdoctoral Fellowship 19-2 FEL-47. (Acknowledgments section) |
| Pseudocode | Yes | Algorithm 1 EC-GP-UCB (Enlarged Confidence GP-UCB) (Page 4) and Algorithm 2 Phased GP Uncertainty Sampling (Page 5) and Algorithm 3 Regret bound balancing [33] (Page 6) |
| Open Source Code | No | The paper does not contain any statements or links indicating that the source code for the described methodology is publicly available (e.g., a GitHub link, an explicit statement of code release, or reference to supplementary materials containing code). |
| Open Datasets | No | The paper is theoretical, designing algorithms and deriving regret bounds. It defines a general compact set of actions D and a black-box reward function f, but it does not mention the use of any specific publicly available datasets for empirical training or evaluation. |
| Dataset Splits | No | The paper is theoretical and does not involve empirical experiments with datasets. Therefore, there are no descriptions of training, validation, or test dataset splits. |
| Hardware Specification | No | The paper is theoretical and does not discuss any empirical experiments, thus there is no mention of specific hardware specifications (e.g., GPU models, CPU types, memory) used for running experiments. |
| Software Dependencies | No | The paper describes algorithms and theoretical frameworks but does not specify any software dependencies with version numbers (e.g., Python version, specific library versions like PyTorch 1.x or TensorFlow 2.x) that would be needed for replication. |
| Experiment Setup | No | The paper is theoretical, presenting mathematical formulations and algorithms. While it defines parameters like B, ฮป, and ฯ as part of its theoretical model, it does not describe an empirical experimental setup with specific hyperparameters (e.g., learning rate, batch size, number of epochs) or system-level training settings. |