We work on building ML models that follow causal principles, such as:

• Principle of independent mechanisms: The causal generative process of a system’s variables is composed of autonomous modules that do not inform or influence each other (Peters et al, Elem. of Causal Inference)
• Reichenbach’s common cause principle: If two random variables X and Y are statistically dependent, then there exists a third variable Z that causally influences both.

These models allow us to compute counterfactuals (e.g. what would have happened if this person had this additional genetic mutation), and to properly model interventions (e.g. give patient X treatment Y).

Our work that was motivated by causal ML:

• Bayesian Image Reconstruction using Deep Generative Models, Marinescu et al, 2021
• Najmeh Mashhadi’s PhD project on studying the composition of multiple ML models.

Further Reading:

• J Peters, D Janzing, B Scholkopf, Elements of Causal Inference.
• D. Castro et al, Causality matters in medical imaging
• N. Pawlowski et al, Deep Structural Causal Models for Tractable Counterfactual Inference