Avoiding Undesired Future with Minimal Cost in Non-Stationary Environments
Authors: Wen-Bo Du, Tian Qin, Tian-Zuo Wang, Zhi-Hua Zhou
NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Theoretical and experimental results validate the effectiveness and efficiency of our method under certain circumstances. We evaluate the proposed AUF-MICNS algorithm on two datasets and focus on four aspects including (a) success frequency for Yt S; (b) average alteration cost; (c) average executing time; and (d) mean square error of the estimated parameters ˆθt. |
| Researcher Affiliation | Academia | Wen-Bo Du, Tian Qin, Tian-Zuo Wang, Zhi-Hua Zhou National Key Laboratory for Novel Software Technology, Nanjing University, China School of Artificial Intelligence, Nanjing University, China {duwb, qint, wangtz, zhouzh}@lamda.nju.edu.cn |
| Pseudocode | Yes | Algorithm 1 OGD-based estimator for ˆβj; Algorithm 2 Online-ensemble-based estimator for ˆβj; Algorithm 3 AUF-MICNS |
| Open Source Code | Yes | Justification: The code and data for the main experimental results are provided in the supplemental material. |
| Open Datasets | Yes | Bermuda Data. This is an ecology dataset that records environment variables in Bermuda [10], and the generation order of variables is available [3]. Justification: The code and data for the main experimental results are provided in the supplemental material. The dataset used in the experiments is also accessible from the link of the reference. |
| Dataset Splits | No | The paper mentions 'The observational dataset size is set to 10.' and 'For each dataset, we repeat experiments with 100 rounds 20 times.', but it does not specify explicit training, validation, or test dataset splits (e.g., percentages, sample counts, or cross-validation folds). |
| Hardware Specification | Yes | The experiments are done by using mac OS Monterey, Apple M1 Pro. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers (e.g., library names with versions). |
| Experiment Setup | Yes | The desired probability is set to τ = 0.7. The observational dataset size is set to 10. The feasible alteration values are [ 3, 3] for centralized P and C, associated with cost coefficients 1.0 and 2.0 respectively since altering P is easier than C. |