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..
Balanced Active Inference
Authors: Boyu Chen, Zhixiang Zhou, Liuhua Peng, Zhonglei Wang
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Various numerical experiments, including regression and classification in both synthetic setups and real data analysis, demonstrate that the proposed algorithm outperforms its alternatives while guaranteeing nominal coverage. Our code is available at: https://github.com/Uninfty/Balanced_Active_Inference |
| Researcher Affiliation | Academia | 1School of Economics, Xiamen University, Xiamen, China 2Shanghai Innovation Institute, Shanghai, China 3School of Mathematics and Statistics, The University of Melbourne, Melbourne, Australia 4Wang Yanan Institute for Studies in Economics, Xiamen University, Xiamen, China |
| Pseudocode | Yes | Algorithm 1 Balanced active inference |
| Open Source Code | Yes | Our code is available at: https://github.com/Uninfty/Balanced_Active_Inference |
| Open Datasets | Yes | Regression datasets include Bike Sharing [32], Communities and Crime [33], Concrete Compressive Strength [34], Energy Efficiency [35], Life Expectancy [36], Superconductivity Data [37], and binary classification datasets including Credit Fraud Detection [38] and Post-election Survey Research [39]. |
| Dataset Splits | No | The numerical analysis proceeds as follows. First, we generate (X, Y ) pairs and randomly split them into training/test sets. |
| Hardware Specification | Yes | All experiments were conducted on a machine equipped with an Intel Xeon Gold 5118 CPU @ 2.30GHz, featuring 12 cores and 24 threads. |
| Software Dependencies | No | All predictive models are obtained by XGBoost [40]. The cube method is implemented by the R package balancesampling via Python s rpy2 interface [41]. |
| Experiment Setup | Yes | For all the datasets, the reported results are based on T = 10 000 Monte Carlo simulations. Following recommendations from the traditional active inference literature [10], we set τ = 0.5 in (6) across all experiments. All predictive models are obtained by XGBoost [40]. |