Amortized Inference of Causal Models via Conditional Fixed-Point Iterations

arXiv stat.ML / 4/6/2026

💬 OpinionIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper proposes an amortized inference framework for Structural Causal Models (SCMs), aiming to infer causal mechanisms without training a separate model for each dataset.
It uses a transformer-based architecture to learn dataset embeddings from observational data, then extends the Fixed-Point Approach (FiP) to perform conditional fixed-point iterations based on those embeddings.
A key capability is generating both observational and interventional data for novel SCMs at inference time without updating model parameters.
Experiments indicate the amortized approach matches per-dataset baselines on both in-distribution and out-of-distribution tasks, and can outperform them in low-data (scarce data) settings.

Abstract

Structural Causal Models (SCMs) offer a principled framework to reason about interventions and support out-of-distribution generalization, which are key goals in scientific discovery. However, the task of learning SCMs from observed data poses formidable challenges, and often requires training a separate model for each dataset. In this work, we propose an amortized inference framework that trains a single model to predict the causal mechanisms of SCMs conditioned on their observational data and causal graph. We first use a transformer-based architecture for amortized learning of dataset embeddings, and then extend the Fixed-Point Approach (FiP) to infer the causal mechanisms conditionally on their dataset embeddings. As a byproduct, our method can generate observational and interventional data from novel SCMs at inference time, without updating parameters. Empirical results show that our amortized procedure performs on par with baselines trained specifically for each dataset on both in and out-of-distribution problems, and also outperforms them in scarce data regimes.