University of Cambridge > Talks.cam > Machine Learning Reading Group @ CUED > Towards Neuro-Causality: Relating Graph Neural Networks to Structural Causal Models

Towards Neuro-Causality: Relating Graph Neural Networks to Structural Causal Models

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Chaochao Lu.

Xia, Lee, Bengio and Bareinboim recently formalized the Causal-Neural Connection in spirit of previously existing work (e.g. Kocaoglu et al. 2017, Ke et al. 2020). This talk will start with an introduction to this arguably new research direction of interest: Neuro-Causality. Thinking of pure Causality as formalized by Judea Pearl in his seminal work, it can be described in terms of a Structural Causal Model (SCM) that carries information on the variables of interest and their mechanistic relations. For most processes of interest the underlying SCM will only be partially observable, thus causal inference tries to leverage any exposed information. Most recently, Zečević, Dhami, Veličković, and Kersting considered the special network type known as Graph Neural Networks (GNN), which act as universal approximators on structured input, for causal learning – thereby suggesting a tighter integration with SCM . For said work, starting from first principles the talk will examine key theoretical results. Finally, the talk will conclude with a perspective on interesting future research directions for neuro-causality.

This talk is part of the Machine Learning Reading Group @ CUED series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2024 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity