Causality Boot Camp
2022-01-18..2022-01-22
Berkeley, United States
as the organisers put it
The boot camp is intended to acquaint program participants with the key themes of the program. It will consist of four days of tutorial presentations from leading experts in the topics of the program.
who it's for
causal inference researchers, machine learning researchers, applied statisticians, methodologists in data science, econometricians, data scientists doing causal analysis, graduate students in statistics
topics
causal graphical models with DAGs; d-separation and conditional independence; do-calculus and identifiability; Shpitser-Pearl ID algorithm; causal discovery with the PC algorithm; score-based search and GES; FCI for latent confounding (PAG/MAG); kernel conditional independence tests (KCI); single-world intervention graphs (SWIGs); potential outcomes and encouragement designs; instrumental-variable and polytope bounds; fixing operation and district decomposition
| title | author | length |
|---|---|---|
| Introduction to Causal Graphical Models: Graphs, d-separation, do-calculus | Spencer Gordon | 01h 06m |
Introduces a probabilistic causal model with hidden U variables, observable Xs, deterministic structural functions and interventions modeled by replacing… | ||
| Introduction to Causal Graphical Models: Graphs, d-separation, do-calculus | Spencer Gordon | 01h 06m |
Defines d-separation precisely (chains, forks, colliders and descendant rules) and proves that d-separation implies conditional independence for all compatible… | ||
| Introduction to Causal Graphical Models: Graphs, d-separation, do-calculus | Spencer Gordon | 01h 01m |
Proves soundness proof by transforming G into a sequence G0…GK: complete ancestor subgraphs, flip edges, and argue each step preserves d-separation with… | ||
| Introduction to Causal Graphical Models: Graphs, d-separation, do-calculus | Spencer Gordon | 01h 08m |
Proves that the three do-calculus rules suffice for deciding identifiability by using an augmented graph lemma that translates d‑separation with intervention… | ||
| Classical Strategies and Concepts in Causal Discovery | Daniel Malinsky | 01h 02m |
Overview of causal discovery methods with concrete claims: PC algorithm uses a staged schedule of conditional independence tests (start with marginals, grow… | ||
| Classical Strategies and Concepts in Causal Discovery | Daniel Malinsky | 01h 06m |
Explains how to run PC and score-based search end to end, with concrete tests (the Fisher Z partial-correlation test for Gaussians, chi-square/G-tests for… | ||
| Unmeasured Confounding and More Recent Developments/Challenges in Causal Discovery | Daniel Malinsky | 01h 02m |
Describes how constraint‑based causal discovery extends to unmeasured confounding by searching PAG/MAG space with the FCI algorithm; shows the… | ||
| Unmeasured Confounding and More Recent Developments/Challenges in Causal Discovery | Daniel Malinsky | 01h 08m |
Covers practical differences between MAGs/PAGs and ADMGs: an edge flagged as visible edges implies no unmeasured confounder between its endpoints and can be… | ||
| An Introduction to Potential Outcomes, DAGs and Single-World Intervention Graphs | Thomas S Richardson | 01h 09m |
Defines potential outcomes as paired counterfactuals per unit, models binary treatment/outcome with four response types (helped, hurt, always/never recover)… | ||
| An Introduction to Potential Outcomes, DAGs and Single-World Intervention Graphs | Thomas S Richardson | 01h 08m |
Shows how a randomized encouragement design with noncompliance (binary Z, X, Y) becomes multiple observational studies, builds one polytope per Z level and… | ||
| An Introduction to Potential Outcomes, DAGs and Single-World Intervention Graphs | Thomas S Richardson | 01h 07m |
Introduces a concrete procedure that turns a DAG into a single-world intervention graph by node-splitting, giving an explicit factorization and a modularity… | ||
| An Introduction to Potential Outcomes, DAGs and Single-World Intervention Graphs | Thomas S Ric... | 01h 03m |
Presents a complete, constructive identification procedure: re-express P(Y|do(X)) as a sum of district-specific intervention kernels, then test each kernel for… | ||
| Introduction to Experimental Design | Chandler Squires | 01h 01m |
Overview of concrete results and assumptions: experimental design targets causal-structure learning under DAG and no-latents assumptions, using DO/knockout… | ||
| Introduction to Experimental Design | Chandler Squires | 54m |
Shows that with no observational data identifying any P‑node DAG needs floor(log2 P)+1 unbounded interventions, while with observational data the count depends… | ||
| Introduction to Experimental Design | Chandler Squires | 01h 04m |
Covers adaptive experiment design under a fixed interventional Markov equivalence class, comparing worst-case, average-case and entropy-based greedy objectives… | ||
| nothing matches | ||