Autoplay
Autocomplete
Speed
Previous Lecture
Complete and Continue
Causal Modeling in Machine Learning Workshop (Alumni)
Lectures for Causal Modeling in Machine Learning Alumni
Introduction
Causality and Probabilistic Graphical Models
Building a Causal Model as a Directed Graph (1:16)
Reasoning about Probability with the DAG (1:46)
Training Causal Probability Distributions on a DAG (2:29)
Causal Markov Kernels and Parameter Modularity (4:32)
Generation and Inference (1:43)
Assessment: Modeling with DAGs
Graph Structure and Conditional Independence
A Computer Science Perspective on DAGs (4:24)
D-Separation (6:25)
Key Graphical Concepts in D-Separation: V-Structures and Markov Blankets
Using D-Separation to Describe Conditional Independence (4:42)
Properties and Assumptions of Causal Models
The Causal Markov Property (3:12)
Assessment - Validating the Markov Property
Faithfulness and Minimality (1:06)
Markov Equivalence (2:24)
Partially Directed Acyclic Graphs and Markov Equivalence (1:51)
Coding Equivalence and PDAGs (1:56)
Causal Sufficiency: How Big Should My Model Be? (2:01)
Latent Variables and Ancestral Graphs
Blending Causal Reasoning and Generative Machine Learning
Models as Symbolic Explanation Generators for Data
Deep Generative Causal Models
Causal Modeling with Variational Autoencoders
Causal Considerations of Discriminative vs. Generative ML (7:16)
Theory of Interventions
Interventions and "No Causation without Manipulation" (3:35)
Modeling Interventions with a DAG
Assessment - Interventions via Graph Mutilation
Structural Interventions and Mechanism
Interventional Sufficiency, Karl Popper, and Falsifiability
Causal Discovery 101: How (Not) to Learn Graphs from Data
Programming Probabilistic Causal Models
From Bayesian Networks to Probabilistic Reasoning Systems (8:20)
Probabilistic Programming Defined (10:19)
Execution, Sampling, and Conditioning (16:39)
PPL Landscape and Deep Probabilistic Programming
Programming Causality (6:26)
Motivating Examples of Causal Probabilistic Programming
Structural Causal Models
Structural Causal Models as Generative Models (3:22)
Asimov and Laplace Explain SCMs (3:52)
Programming SCMs (2:20)
Interventions on SCMs (1:47)
Independence of Mechanism (2:55)
SCM Assessment
Applied Causal Inference; Identification and Estimation of Causal Effects from Data
Motivating Causal Effect Inference (6:03)
Causal Effects and Common Cause (6:55)
Simpson's Paradox (5:29)
Defining Causal Inference with Interventions (6:44)
Recap on Causal Generative Reasoning Systems (4:08)
Identifiability for Causal Queries (2:29)
"Do"-Calculus: Identification without a Generative Model (2:07)
A Closer Look at the Do-Calculus
Valid Adjustment Sets and the Adjustment Formula (4:57)
Front-door Adjustment (1:50)
Assessment - Do-Calculus and Valid Adjustment
Statistical Methods for Causal Inference
Instrumental Variables (10:47)
Causality and "The Bitter Lesson"
Propensity Scores and Matching (3:58)
Inverse Probability Reweighting Methods (4:17)
You can use Causal Generative Models for Causal Inference (10:16)
Being Bayesian about model parameters
Assessment - Statistical Methods for Causal Inference
Potential Outcomes and their Contrasts to Causal Graphical Models
Potential Outcomes Framework and Assumptions (8:11)
SCMs vs PO; Heterogeneous Populations vs Individual Treatment Effects
"So then are SCM's better than PO?" and other FAQ
G-Methods, G-Estimation, and Time-varying Treatments
Algorithmic Structural Counterfactuals
Introduction to Counterfactual Reasoning (3:46)
The Twin-World Counterfactual Inference Algorithm (9:18)
Assessment - Counterfactuals
Mediation: An Algorithmic Bias Case Study
Assessment - Mediation
Single World Counterfactuals & Effect of Treatment on the Treated
Probabilities of Neccessity and Sufficiency
Picking an SCM
Assessment - Necessity and Sufficiency
Wrapping Up
You did it! Next steps.
Simpson's Paradox
Lecture content locked
If you're already enrolled,
you'll need to login
.
Enroll in Course to Unlock