Ordering-Based Causal Discovery with Reinforcement Learning
Authors: Xiaoqiang Wang, Yali Du, Shengyu Zhu, Liangjun Ke, Zhitang Chen, Jianye Hao, Jun Wang
IJCAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We conduct comparative experiments on synthetic and real data sets to validate the performance of the proposed methods. |
| Researcher Affiliation | Collaboration | 1State Key Laboratory for Manufacturing Systems Engineering, School of Automation Science and Engineering, Xi an Jiaotong University 2University College London 3Huawei Noah s Ark Lab 4College of Intelligence and Computing, Tianjin University |
| Pseudocode | Yes | Algorithm 1 Causal discovery with Ordering-based RL. |
| Open Source Code | Yes | An implementation has been made available at https://github. com/huawei-noah/trustworthy AI/tree/master/gcastle. |
| Open Datasets | Yes | The Sachs data set [Sachs et al., 2005], with 11-node and 17-edge true graph, is widely used for research on graphical models. |
| Dataset Splits | No | The paper discusses mini-batch training and generating synthetic data, but it does not specify explicit train/validation/test splits with percentages or absolute sample counts for the datasets used in experiments. It mentions using 'different sample numbers' but no formal split methodology. |
| Hardware Specification | No | The paper mentions running times for different methods (e.g., 'CORL-1 and CORL-2 7 hours against Gra N-DAG and DAG-GNN 4 hours for 100-node graphs') but does not provide any specific hardware details such as GPU models, CPU specifications, or memory. |
| Software Dependencies | No | The paper mentions using specific methods and models (e.g., 'linear regression', 'CAM pruning', 'policy gradient', 'critic network') and states they use 'original implementations' for baselines, but it does not provide specific version numbers for any software libraries or environments (e.g., Python, PyTorch, TensorFlow, scikit-learn versions). |
| Experiment Setup | Yes | For variable selection, we set the thresholding as 0.3 and apply it to the estimated coefficients, as similarly used by [Zheng et al., 2018; Zhu et al., 2020]. For the non-linear model, we adopt the CAM pruning used by [Lachapelle et al., 2020]. We also set a maximum time limit of 15 hours for all the methods for fair comparison and only graphs with up to 30 nodes are considered here. |