Machine Learning Trick of the Day: Instrumental Thinking | Hacker News Reader

Machine Learning Trick of the Day: Instrumental Thinking

blog.shakirm.com

125 pointsghosthamlet10 comments

Shakir is a force of nature. He is one of the organizers of the Deep Learning Indaba - an incredible effort to strengthen machine learning in Africa. Unfortunately the 2018 videos aren't out yet (Shakir gave an amazing talk on deep learning fundamentals and some of the tricks from his blog) but everyone should be aware of Shakir and the Deep Learning Indaba: http://www.deeplearningindaba.com/

Seems related to noisy-input Gaussian Process [1] that I applied in my thesis with moderate success some years ago.

[1] http://mlg.eng.cam.ac.uk/pub/pdf/MchRas11.pdf

Excellent post about potential flaws of your typical linear regression model and how to address with causal methods. As a self-taught practitioner, these posts are priceless in letting me know what I don't know...

Does the machine learning field have different nomenclature for this sort of thing than, say, econometrics?

I recognize all the math, but the verbiage used is different.

I'm a trained a econometrician and a data scientist. Have you found people to be apprehensive with you since 'our' verbiage is different?

Is that what I asked?

I'm asking about why the usage of the mathematics that you and I are apparently both familiar with could be described in different ways depending on which field is referencing it. This is a linguistic question - not a mathematical one.

No need to be rude.

Is that what they asked?

Nomenclature is different in different fields all the time, or even different areas of the same field. It all depends on what practitioners feel is most intuitive.

No need to call someone rude for not answering your question and asking their own instead.

What was rude? A question?

Each field seems to have its own nomenclature. I recognize 'instrumental variables' from a (bio-)statistics course, so I think that's a pure statistics term. Larry Wasserman has a vocabulary chart in 'All of Statistics' (there's a PDF I won't link to) that lists terms differing between statistics and computer science. For example he writes that 'using data to estimate an unknown quantity' is called 'estimation' in statistics and 'learning' in computer science.

Shakir is a fantastic technical writer. I love that he also traces modern ideas back to their older roots and includes the references.

Reply on news.ycombinator.com