A survey paper on causal representation learning and counterfactual generation

We are pleased to share the survey paper:

From Identifiable Causal Representations to Controllable Counterfactual Generation: A Survey on Causal Generative Modeling, Aneesh Komanduri, Xintao Wu, Yongkai Wu, and Feng Chen. https://arxiv.org/pdf/2310.11011.pdf

Deep generative models have shown tremendous success. However, their shortcomings include lack of explainability, the tendency to induce spurious correlations, and poor out-of-distribution extrapolation. To remedy such challenges, one can incorporate the theory of causality (e.g., structural causal models) in deep generative modeling. Structural causal models (SCMs) describe data-generating processes and model complex causal relationships and mechanisms among variables in a system, and offer several beneficial properties to deep generative models, such as distribution shift robustness, fairness, and interoperability. We provide a technical survey on causal generative modeling categorized into causal representation learning and controllable counterfactual generation methods. We focus on fundamental theory, formulations, drawbacks, datasets, metrics, and applications of causal generative models in fairness, privacy, out-of-distribution generalization, and precision medicine. We also discuss open problems and fruitful research directions for future work in the field.