Causality Learning With Wasserstein Generative Adversarial Networks
This work addresses the problem of improving causal structure learning for researchers in machine learning, but it is incremental as it builds on existing continuous optimization frameworks by incorporating Wasserstein distance.
The paper tackles the challenge of causal structure learning from data by proposing DAG-WGAN, a model that integrates Wasserstein distance with an acyclicity constraint in an auto-encoder architecture, and it performs better with high cardinality data in experiments.
Conventional methods for causal structure learning from data face significant challenges due to combinatorial search space. Recently, the problem has been formulated into a continuous optimization framework with an acyclicity constraint to learn Directed Acyclic Graphs (DAGs). Such a framework allows the utilization of deep generative models for causal structure learning to better capture the relations between data sample distributions and DAGs. However, so far no study has experimented with the use of Wasserstein distance in the context of causal structure learning. Our model named DAG-WGAN combines the Wasserstein-based adversarial loss with an acyclicity constraint in an auto-encoder architecture. It simultaneously learns causal structures while improving its data generation capability. We compare the performance of DAG-WGAN with other models that do not involve the Wasserstein metric in order to identify its contribution to causal structure learning. Our model performs better with high cardinality data according to our experiments.