Re: [AIRG] Causal inference in statistics 3/13

[Aubrey Barnard <barnard@xxxxxxxxxxx>]

AIRG,

I would like to remind you all that this afternoon Tim Huegerich will be 
presenting an overview of causal inference in statistics. Causal 
reasoning is one of the core pieces of intelligence, and so will 
continue to be important in advancing AI, regardless of the successes of 
deep nets. (AGI will need to reason about causality as well as model 
complex functions.)

Causality can be an intimidating / expansive topic, but all you really 
need to know is that you do causal reasoning with Bayes nets (or similar 
functional models). Join us this afternoon to see how it's done!

4pm, CS 3310
http://dx.doi.org/10.1214/09-SS057

Aubrey


On 3/8/19 12:36 PM, Tim Huegerich wrote:
> AIRGers,
> 
> On Wednesday, March 13, at 4pm,
> in CS 3310,
> Tim Huegerich presents:
> "Causal inference inÂstatistics: An overview"
> by Judea Pearl
> /Statistics Surveys/ 3, 2009
> http://dx.doi.org/10.1214/09-SS057
> (Let me know if you have any access trouble.)
> 
> I will be presenting an introduction to causalÂinference in statistics,
> especially as it pertains to observational dataÂand not randomized
> controlled experiments. Causality in observationalÂdata is quite
> important (because experiments are not always feasible orÂethical),
> and it applies to many fields including economics and medicine.
> 
> The paper is long and I don't expect you to read all of it. (And it
> would be impossible to cover any substantial portion in a single AIRG
> session.) Here is what I suggest you read, in order of priority:
> 
> 1. Abstract and Sections 1-2 (5.5 pages) for context.
> 2. As much of Section 3 as you can, but mainly up through Section 3.3.2
> (~15 pages). This is the core technical material and where I will focus
> my presentation.
> 3. Introduction to Section 4 (0.5 pages). Explanation of the classic
> potential outcome framework.
> 4. Any more is further study for you and I do not plan to cover it.
> 
> See you next week.
> 
> Tim
> 
> p.s. Aubrey presented this same paper back in 2014, and credit for the 
> summary above goes to him--I have copied the summary he used almost 
> verbatim.
> 
> _______________________________________________
> AIRG mailing list
> AIRG@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/airg
>