Gradient-based Causal Structure Learning with Normalizing Flow
This work addresses causal inference for researchers and practitioners in machine learning, offering a more efficient method for structure learning, though it appears incremental by building on existing techniques like NOTEARS and normalizing flows.
The paper tackles the problem of learning causal structures from observational data by proposing DAG-NF, a score-based normalizing flow method that uses Jacobian matrices to infer causal relationships, and it extends NOTEARS to enforce acyclicity constraints, significantly reducing computational complexity.
In this paper, we propose a score-based normalizing flow method called DAG-NF to learn dependencies of input observation data. Inspired by Grad-CAM in computer vision, we use jacobian matrix of output on input as causal relationships and this method can be generalized to any neural networks especially for flow-based generative neural networks such as Masked Autoregressive Flow(MAF) and Continuous Normalizing Flow(CNF) which compute the log likelihood loss and divergence of distribution of input data and target distribution. This method extends NOTEARS which enforces a important acylicity constraint on continuous adjacency matrix of graph nodes and significantly reduce the computational complexity of search space of graph.