Choosing Between Dags Can Be Hard

on YouTube ↗11m

" summary

Two DAGs that are observationally equivalent sit on different algebraic constraints in correlation space: one is a plane enforcing marginal independence, the other is a hyperbolic parabola encoding conditional independence, and they intersect on the axes and at the origin. Near the origin the distance between models scales like 1/N, so effects of order 1/√N are indistinguishable and strong-faithfulness bounds still fail. A decision rule uses a likelihood-ratio threshold and otherwise allows the answer report a collection of models.

AI-written summary · about

" chapters (8)

1. 0:00Introduction
2. 0:15Richard Go
3. 1:38Geometric View
4. 2:35Scoring Approach
5. 2:50Ratio Test
6. 4:08Theory
7. 5:40Decision Rule
8. 7:07Whats Next

" related talks

• 44mMarginal DAG Models Equivalent to Another DAGRobin Evans · Quantum Physics and Statistical Causal Models (Simons Institute workshop, 2022)
• 01h 06mIntroduction to Causal Graphical Models: Graphs, d-separation, do-calculusSpencer Gordon · Causality Boot Camp
• 52mNew Algorithms For The Graph Alignment ProblemMarc Lelarge · Algorithmic Advances for Statistical Inference with Combinatorial Structure
• 01h 30mHigh-Dimensional Statistics IMartin Wainwright · Big Data Boot Camp 2013
• 48mComputational-Statistical Tradeoffs in Robust High-Dimensional EstimationIlias Diakonikolas · Robust and High-Dimensional Statistics (Simons Institute workshop, 2018)
• 01h 08mIntroduction to Causal Graphical Models: Graphs, d-separation, do-calculusSpencer Gordon · Causality Boot Camp
• 48mThe Approximate Implication Problem For Probabilistic Graphical ModelsBatya Kenig · Algorithmic Aspects of Causal Inference 2022
• 01h 08mData Driven Optimization Models and AlgorithmsYinyu Ye · Optimization, Statistics and Uncertainty