1/4 Broadly, because estimating a cond. prob. P(Y | X) is easier than a joint one, P(Y, X, Z), when X & Z are high-dimensional. Indeed, this observation led to (non-causal) graphical models in the 1980s, including Bayesian nets (ie, non-causal DAGs), Markov random fields & so on.
2/4 The goal in this journey was to find parsimonious encoding of distributions that may lead to efficient learning-inference (e.g., belief propagation). There is the whole UAI community, which is now permeated throughout other ML conferences, focused on this class of algorithms.
3/4 This was insufficient for causal reasoning, as noted in the early 1990's & discussed in Sec. 1.4 here causalai.net/r60.pdf (e.g., see ex. 10). It has been a quite traumatic transition, as noted by @yudapearl , see quote found in p. 30:
4/4 Why are BNs not enough? What is missing? Additional assumptions of modularity & better semantics for the arrows in the model are needed, which has been done by Pearl (Causality, p. 23), and refined in causalai.net/r384-lnai.pdf , and generalized R-60 (Def. 12/16).

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Elias Bareinboim

Elias Bareinboim Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @eliasbareinboim

8 Jul 19
1/5 The Backdoors, Frontdoors, IVs & Duplexes of the world are just specific identification (ID) strategies that should come later, in other words, the premise of the discussion seems ‘odd’ to me. First, one should model the phenomenon under investigation using its best science,
2/5 regardless of what her/his favorite ID strategies are. After having a model of reality (e.g., a causal DAG), she/he can go on and discuss what ID strategy is more suitable for answering the specific query. As I said, I find it a bit curious that CS folks &
3/5 methodologists are clearly against starting w/ any procedure (e.g., backdoor, g-computation), while some empirical scientists seem to be thinking first about these procedures, conflating ID strategy & model construction. In summary: model reality first, ID & estimation later.
Read 5 tweets
9 Jun 19
1/n Dear friends, I am pleased to announce that after four productive years here at Purdue, I decided to move to Columbia University, starting on this July 1st. I am excited about the new possibilities and adventures that we’ll have in causality-land in the next years, both in
2/n terms of the science as well as the applications of causal inference (CI).

I want to share a Q I have been asked more frequently in the last years, which is whether I believe CI is done, at least in terms of basic principles, and if all that is needed are applications.
3/n I couldn't disagree more with the statement. (For details, check my talk Judea alluded, which is available @ cs.columbia.edu/streaming/2019…).

I don't think anyone would deem molecular biology "solved", and ready to apply (e.g., to solve cancer) when Watson and Crick published...
Read 8 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!