Amortized learning of neural causal representations
This work addresses scalability and knowledge transfer issues in causal modeling, which is incremental as it builds on existing neural network methods.
The paper tackles the challenge of learning causal models that scale poorly with variable count and cannot reuse prior knowledge, by introducing causal relational networks (CRN) that use neural networks for continuous representations, achieving high accuracy and quick adaptation on synthetic data.
Causal models can compactly and efficiently encode the data-generating process under all interventions and hence may generalize better under changes in distribution. These models are often represented as Bayesian networks and learning them scales poorly with the number of variables. Moreover, these approaches cannot leverage previously learned knowledge to help with learning new causal models. In order to tackle these challenges, we represent a novel algorithm called \textit{causal relational networks} (CRN) for learning causal models using neural networks. The CRN represent causal models using continuous representations and hence could scale much better with the number of variables. These models also take in previously learned information to facilitate learning of new causal models. Finally, we propose a decoding-based metric to evaluate causal models with continuous representations. We test our method on synthetic data achieving high accuracy and quick adaptation to previously unseen causal models.