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 Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
Reinforcement Learning with Imperfect Transition Predictions: A Bellman-Jensen Approach
Authors: Chenbei Lu, Zaiwei Chen, Tongxin Li, Chenye Wu, Adam Wierman
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We validate our theory and algorithm on synthetic MDPs and a real-world wind energy storage control problem. ... Section 6 then provides empirical validation on a wind energy storage problem. |
| Researcher Affiliation | Academia | Chenbei Lu Institute for Interdisciplinary Information Sciences Tsinghua University Zaiwei Chen Edwardson School of Industrial Engineering Purdue University Tongxin Li School of Data Science The Chinese University of Hong Kong (Shenzhen) Chenye Wu School of Science and Engineering The Chinese University of Hong Kong (Shenzhen) Adam Wierman Computing & Mathematical Sciences Caltech |
| Pseudocode | Yes | Algorithm 1 summarizes the BOLA procedure. Algorithm 1 BOLA: Bayesian Offline Learning with Online Adaptation |
| Open Source Code | Yes | We included code and data in supplementary files. |
| Open Datasets | Yes | In the numerical study, we utilized the California aggregate wind power generation dataset from CAISO [60] containing predicted and real wind power generation data with a 5-minute resolution spanning from January 2020 to December 2020. ... [60] California ISO. Electricity Price Data. https://www.energyonline.com/Data/, 2021. Online; accessed on 20 December 2022. |
| Dataset Splits | No | In the numerical study, we utilized the California aggregate wind power generation dataset from CAISO [60] containing predicted and real wind power generation data with a 5-minute resolution spanning from January 2020 to December 2020. (This describes the dataset's period but not how it was explicitly split for training, validation, or testing.) |
| Hardware Specification | No | The paper does not explicitly state specific hardware details such as GPU/CPU models, processor types, or memory amounts used for running the experiments. The NeurIPS checklist indicates this information is in the Appendix, but the provided text for Appendices G and H does not contain these specifics. |
| Software Dependencies | No | The paper does not explicitly mention specific versions of software dependencies such as programming languages, libraries, or frameworks (e.g., Python 3.x, PyTorch 1.x). |
| Experiment Setup | Yes | We set C = 10 k Wh, γ = 0.95. The discretization levels of p, w and So C are set to be 10, 10, 21, respectively. The action set includes 9 discretized choices ranging from charging 2 KWh to discharging 2 KWh. |