Some Staged Tree Models For Learning From Interventions

on YouTube ↗29m

" summary

Defines a restricted family of colored probability trees called CS-trees and proves a global Markov-type property: distributions in the model correspond exactly to being Markov to a sequence of context graphs built from the tree’s minimal contexts. Uses stage tree / chained event graph parametrization, an ordering assumption (fixed total order, no truncation), and an augmentation for soft interventions that requires complete intervention targets. Results give a Verma–Pearl–style structural equivalence criterion and outline where the ordering and completeness restrictions block straightforward generalization.

AI-written summary · about

" chapters (8)

1. 0:00Start
2. 2:17Probabilistic Theory of Causality
3. 3:36Stage Tree Models
4. 3:46Stage Tree Model
5. 6:33Models That Allow for Stochastic Soft Interventions
6. 8:36Interpretability
7. 15:54Specialization and Absorption
8. 17:03Minimal Contexts

" related talks

• 24mA Friendly Introduction To CausalityAlex Dimakis · MLconf SF 2016
• 35mUnderstanding, Simplicity, And Decision TreesBin Yu · Data-Driven Decision Processes Boot Camp 2022
• 01h 04mIntroduction to Experimental DesignChandler Squires · Causality Boot Camp
• 01h 01mIntroduction to Experimental DesignChandler Squires · Causality Boot Camp
• 33mLearning a Tree-Structured Ising Model in Order to Make PredictionsGuy Bresler · Approximate Counting, Markov Chains and Phase Transitions
• 01h 06mIntroduction to Causal Graphical Models: Graphs, d-separation, do-calculusSpencer Gordon · Causality Boot Camp
• 01h 08mAn Introduction to Potential Outcomes, DAGs and Single-World Intervention GraphsThomas S Richardson · Causality Boot Camp
• 01h 07mAn Introduction to Potential Outcomes, DAGs and Single-World Intervention GraphsThomas S Richardson · Causality Boot Camp