Ordering-Based Causal Discovery for Linear and Nonlinear Relations

Authors: Zhuopeng Xu, Yujie Li, Cheng Liu, Ning Gui

NeurIPS 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate that our proposed solutions outperform state-of-the-art baselines on synthetic data with varying ratios of linear and nonlinear relations. The results obtained from real-world data also support the competitiveness of Ca PS.
Researcher Affiliation Academia Zhuopeng Xu Yujie Li Cheng Liu Ning Gui School of Computer Science and Engineering Central South University {xuzhuopeng, yujieli}@csu.edu.cn, {liuchengstudy, ninggui}@gmail.com
Pseudocode Yes Algorithm 1 Ordering and Computing parent score (page 5) and Algorithm 2 Post-processing (Appendix B, page 10).
Open Source Code Yes Code and datasets are available at https://github.com/E2real/Ca PS.
Open Datasets Yes Code and datasets are available at https://github.com/E2real/Ca PS. (Abstract) Synthetic data are created using the Erdös-Rényi (ER) [30] or Scale-Free (SF) models[31]... Real dataset contains a protein expression dataset Sachs [1] and a pseudoreal transport network dataset Syntern [32].
Dataset Splits No The paper defines evaluation metrics but does not specify explicit training, validation, or test dataset splits (e.g., percentages or sample counts) for the datasets used in the experiments.
Hardware Specification Yes All experiments were run on EPYC 7552*2 with 512G memory and NVIDIA RTX 4090 32GB.
Software Dependencies No The paper mentions methods and tools used (e.g., 'CAM pruning'), but it does not specify versions for software dependencies like programming languages or specific libraries.
Experiment Setup Yes The only hyperparameter of Ca PS was rigor λ, which we set to λ = 50 for all datasets to avoid any dataset-specific tuning.