Foundations of causal inference and its impacts on machine learning webinar

[@marcelobeckmann9552]

Amazing work, and this for sure will change the world! The Nobel Prize in Economics from 2021 came to a research in Causation, but thanks to initiatives like this we don't need to be a Nobel Laureated to work with Causation. Its a pity we can't give more than one like for this video.

[@anirbanchakraborty7319]

Having just completed the Book Of Why, I was convinced of its enormous importance. But all the while had this feeling that how can causal inference be of use in fuzzy problems where we have absolutely no intuitive idea of a causal structure. In other words, how can causal inference be of help to machine learning engineers and not just to social science experts. This video gives clear indications towards that end. Overall enlightening.

[@idkwiadfr]

Hi, I am starting my master thesis on the same topic, could you please help me find best resources to get going with the topic. It will be of great help. Thank you

[@markusloecherNJ]

Very clear presentation, thanks !! On the identification slide (e.g. time=22:19) might there be a mistake? Should the text not say P(Y | do(T)) instead of P(T | do(Y)) ?

[@tilli7500]

3 qs for amit sharma 1- how do you deal with the encoder bias,if we could call it that, i e what if domain experts introduce bias in the process?? wouldn't it be more harmful as being active rather than being passive(learnt from a dataset).??? 2- 1:09:00 consider a classical ml method decision tree classifier,don't we beforehand know which features are more "important" ,using gini??? 3- how can one join ms r india????is a phd/iit tag mandatory????

[@wy2528]

This is really interesting and thank you for the presentation

[@Sreejith284770]

Do Why is a wonderful tool. Thanks for making it open source. It was a very good presentation.

[@idkwiadfr]

Hi, I am starting my master thesis on the same topic, could you please help me find best resources to get going with the topic. It will be of great help. Thank you

[@qiguosun129]

Amazing work, we can't give more than one like for this video!

[@xdxn2010]

18:31, if graph cannot be learned from data alone, what does causal discovery do?

[@user-wr4yl7tx3w]

this is a really great presentation. I noticed that the slides are not available though.

[@gabrielwong1991]

As an econometrican seems to me this is related to Instrumental Variable?

[@user-wr4yl7tx3w]

Can you have contrasting causal graphs leading to opposing causal inferences on Y? Given a data set, is the causal graph unique?

[@alperensayar9679]

very exciting thing. thanks for clear expl.

[@julianmurillo5965]

Love love love this kind of content.

[@jmarkinman]

Fantastic work. But seems difficult to scale.

[@প্রীতমবিশ্বাস]

Wow could anybody suggest me some research papers!

[@OldEarthWisdom]

So, what you are saying is that in the future we will be able to include the effect everything has on the planet. Say, for example, if a company convinces people to buy their new shows we will be able to see how many workers in Bangladesh will need to work 12 hours a day and how their children will live without parents, and how many trees will be cut down in Brazil and how much oil it takes to ship these shoes around the world. Sounds good to me.

[@michaelreeves1147]

No insight.

[@AnimeshSharma1977]

Will do IT for sure! Bummer that kzbin.info/www/bejne/gnKvl4SJqbSBe8U , wondering why "domain knowledge" cannot be filled in with information, isn't it data too?