Tata Institute of Fundamental Research
Uncovering Causality
Student Seminar
Speaker:
Ronak M. Soni (Tata Institute of Fundamental Research
Department of Theoretical Physics
Homi BHabha Road
Navy Nagar
Mumbai 400005)
Organiser:
Gowtham Raghunath Kurri
Date:
Friday, 31 Oct 2014, 14:00 to 15:30
Venue:
D-405 (D-Block Seminar Room)
(Scan to add to calendar)
Abstract:
Abstract: Correlations, as you know, is not causation; in fact, a vast majority of correlations aren't causal. However, human progress has happened not because someone understood that the banging of a hammer was correlated with the burial of a nail but because someone understood that the banging of a hammer caused the burial of the nail; we are interested in doing things and having those actions cause desirable consequences.
Earlier, the only way to abstract causation from data was randomised controlled trials - and those are often broken by confounders. In the past couple of decades, Judea Pearl and his collaborators have begun uncovering causal relationships as properties of particular conditionalisations of joint probability distributions - allowing us to infer causal relationships from data not found using RCTs. And as a bonus, this method of analysis is also robust to identified confounders.
I'll be discussing the general ideas of this analysis, and maybe even some simple examples.
| Speaker: | Ronak M. Soni (Tata Institute of Fundamental Research Department of Theoretical Physics Homi BHabha Road Navy Nagar Mumbai 400005) |
| Organiser: | Gowtham Raghunath Kurri |
| Date: | Friday, 31 Oct 2014, 14:00 to 15:30 |
| Venue: | D-405 (D-Block Seminar Room) |
(Scan to add to calendar)
Abstract:
Abstract: Correlations, as you know, is not causation; in fact, a vast majority of correlations aren't causal. However, human progress has happened not because someone understood that the banging of a hammer was correlated with the burial of a nail but because someone understood that the banging of a hammer caused the burial of the nail; we are interested in doing things and having those actions cause desirable consequences.
Earlier, the only way to abstract causation from data was randomised controlled trials - and those are often broken by confounders. In the past couple of decades, Judea Pearl and his collaborators have begun uncovering causal relationships as properties of particular conditionalisations of joint probability distributions - allowing us to infer causal relationships from data not found using RCTs. And as a bonus, this method of analysis is also robust to identified confounders.
I'll be discussing the general ideas of this analysis, and maybe even some simple examples.
Earlier, the only way to abstract causation from data was randomised controlled trials - and those are often broken by confounders. In the past couple of decades, Judea Pearl and his collaborators have begun uncovering causal relationships as properties of particular conditionalisations of joint probability distributions - allowing us to infer causal relationships from data not found using RCTs. And as a bonus, this method of analysis is also robust to identified confounders.
I'll be discussing the general ideas of this analysis, and maybe even some simple examples.