Causal Discovery
Causal discovery aims to infer cause-and-effect relationships from data, moving beyond simple correlations to understand underlying mechanisms. Current research emphasizes developing algorithms and models, including constraint-based methods, score-matching techniques, and those leveraging neural networks (like graph neural networks and normalizing flows), to efficiently handle high-dimensional data, time series, and latent variables, often incorporating expert knowledge or interventional data to improve accuracy. This field is crucial for advancing scientific understanding across diverse domains, from biology and healthcare to climate science and robotics, by enabling more accurate modeling, prediction, and intervention design. Improved causal discovery methods are leading to more reliable insights and more effective decision-making in complex systems.
Papers
A Causal Ordering Prior for Unsupervised Representation Learning
Avinash Kori, Pedro Sanchez, Konstantinos Vilouras, Ben Glocker, Sotirios A. Tsaftaris
Benchmarking Bayesian Causal Discovery Methods for Downstream Treatment Effect Estimation
Chris Chinenye Emezue, Alexandre Drouin, Tristan Deleu, Stefan Bauer, Yoshua Bengio
Neuro-Causal Factor Analysis
Alex Markham, Mingyu Liu, Bryon Aragam, Liam Solus
Causal discovery for time series with constraint-based model and PMIME measure
Antonin Arsac, Aurore Lomet, Jean-Philippe Poli
Causal Discovery with Latent Confounders Based on Higher-Order Cumulants
Ruichu Cai, Zhiyi Huang, Wei Chen, Zhifeng Hao, Kun Zhang