I see what you're saying and upon re-reading I think it's ambiguous which SME is being referenced. I guess it depends on who the audience is, and I think we both assume the audience is a non-ML SME.
the point here - at least as I read it - is to help the non-ML-expert folks get started applying Deep Learning.
If that is in fact the case, I would argue that this document is not really giving that user the best starting point - though they get points for trying.
As a practitioner perhaps I am biased, and our applications are on the boundaries of solved problems (though include some reasonably solved CV solutions) - however I would argue that unless your application is strictly the simplest and can use off the shelf solutions, for example simply implementing Google Cloud Vision API or the Microsoft Cognitive Computing API - you're going to need someone with years of study/practice with ML to get to a good outcome.
It is on the face of it a pretty weird phrase ("subject matter"), but I've always seen it used to mean what mindcrime means. The phrase is also used in general non-ML specific programming (eg among a team of people building a mobile app for a bank, the "subject matter expert" might be a non-programmer who has years of experience working for banks).
It's also disambiguated by the author saying "subject matter expert for your application" (application being an application of machine learning, eg to radiology, geology...), and saying "subject matter experts in their application but deep learning novices" in the paper's abstract.
I've seen dozens of examples of people with less than a year of ML experience get great outcomes from deep learning. It's no longer true that this is such an exclusive field.
the point here - at least as I read it - is to help the non-ML-expert folks get started applying Deep Learning.
If that is in fact the case, I would argue that this document is not really giving that user the best starting point - though they get points for trying.
As a practitioner perhaps I am biased, and our applications are on the boundaries of solved problems (though include some reasonably solved CV solutions) - however I would argue that unless your application is strictly the simplest and can use off the shelf solutions, for example simply implementing Google Cloud Vision API or the Microsoft Cognitive Computing API - you're going to need someone with years of study/practice with ML to get to a good outcome.