Causal Identification: Selection Bias and Latent Variables
Event details
| Date | 21.08.2023 |
| Hour | 15:00 › 17:00 |
| Speaker | Amir Aboueimehrizi |
| Location | |
| Category | Conferences - Seminars |
| Event Language | English |
EDIC candidacy exam
Exam president: Prof. Lenka Zdeborová
Thesis advisor: Prof. Negar Kiyavash
Thesis co-advisor: Prof. Matthias Grossglauser
Co-examiner: Prof. Yanina Shkel
Abstract
Causal inference aims to identify the effects of some
variables on others within a system. Depending on the objective
and available data, diverse scenarios can emerge. This report
presents an overview of three distinct problems: the identification
of causal effects (1) on the entire population from population data,
(2) on a specific sub-population from observational data of the
population; and (3) on the entire population from observational
data of only a sub-population, where unobservable variables might
exist in all setting.
Furthermore, we discuss our ongoing research, identifying
causal effects on a sub-population using the observational
distribution of that sub-population.
Background papers
1- Identification of Joint Interventional Distributions in Recursive Semi-Markovian Causal Models
2- Identification of Conditional Interventional Distributions
3- Recovering Causal Effects from Selection Bias
Exam president: Prof. Lenka Zdeborová
Thesis advisor: Prof. Negar Kiyavash
Thesis co-advisor: Prof. Matthias Grossglauser
Co-examiner: Prof. Yanina Shkel
Abstract
Causal inference aims to identify the effects of some
variables on others within a system. Depending on the objective
and available data, diverse scenarios can emerge. This report
presents an overview of three distinct problems: the identification
of causal effects (1) on the entire population from population data,
(2) on a specific sub-population from observational data of the
population; and (3) on the entire population from observational
data of only a sub-population, where unobservable variables might
exist in all setting.
Furthermore, we discuss our ongoing research, identifying
causal effects on a sub-population using the observational
distribution of that sub-population.
Background papers
1- Identification of Joint Interventional Distributions in Recursive Semi-Markovian Causal Models
2- Identification of Conditional Interventional Distributions
3- Recovering Causal Effects from Selection Bias
Practical information
- General public
- Free