The official source code: On the Role of Entropy-Based Loss for Learning Causal Structure with Continuous Optimization URL
Please run linear_ent.py:
python linear_ent.py
The demo is to run our method in linear Uniform data with 5 nodes.
Please run nonlinear_ent.py
python nonlinear_ent.py
The demo is to run our method in nonlinear Uniform data with 10 nodes.
We greatly thank the implementation code of xunzheng at https://github.com/xunzheng/notears. Our code is based on the code of this implementation.
We also thank the sachs dataset from https://github.com/kurowasan/GraN-DAG/blob/master/data/sachs.zip
If you find our codes useful, please cite:
W. Chen, J. Qiao, R. Cai and Z. Hao, "On the Role of Entropy-Based Loss for Learning Causal Structure With Continuous Optimization," in IEEE Transactions on Neural Networks and Learning Systems, doi: 10.1109/TNNLS.2023.3327357. keywords: {Optimization;Mathematical models;Computational modeling;Additive noise;Task analysis;Random variables;Nonlinear systems;Acyclicity constraint;causal discovery;entropy-based loss;least-square loss},