Interventional Sum-Product Networks: Causal Inference with Tractable Probabilistic ModelsDownload PDF

May 21, 2021 (edited Oct 29, 2021)NeurIPS 2021 PosterReaders: Everyone
  • Keywords: Interventions, Causality, Tractablity, Probabilistic Models
  • TL;DR: We consider the problem of learning interventional distributions (i.e., answering causal queries) with tractable probabilistic models (gated SPN).
  • Abstract: While probabilistic models are an important tool for studying causality, doing so suffers from the intractability of inference. As a step towards tractable causal models, we consider the problem of learning interventional distributions using sum-product networks (SPNs) that are over-parameterized by gate functions, e.g., neural networks. Providing an arbitrarily intervened causal graph as input, effectively subsuming Pearl's do-operator, the gate function predicts the parameters of the SPN. The resulting interventional SPNs are motivated and illustrated by a structural causal model themed around personal health. Our empirical evaluation against competing methods from both generative and causal modelling demonstrates that interventional SPNs indeed are both expressive and causally adequate.
  • Supplementary Material: pdf
  • Code Of Conduct: I certify that all co-authors of this work have read and commit to adhering to the NeurIPS Statement on Ethics, Fairness, Inclusivity, and Code of Conduct.
  • Code:
9 Replies