This article is terrible. You can’t make a case by putting a bunch of unsupported assertions into section-heading fonts and then just filling in paragraphs.
This reads like a desperate business person wrote it, who wishes that one full-stack set of drives made sense and coexisted in a single person to make that labor cheaper and more commidity, despite the reality that it’s simply not true.
The person who spent the time to master web service frameworks, query languages and product engineering necessarily did not also master professional level knowledge of deep learning or MCMC sampling or natural language processing.
The two types of people need to coexist and work symbiotically, but it’s just asinine wishful thinking to pretend like they are the same person, let alone to write a baseless essay full of assertions that if they aren’t the same person it somehow results in first principles economic inefficiency.
You'll have to do better than an ad hominem + "the opposite is true." Author is Chief Algorithms Officer at Stitchfix and former VP Data Science & Engineering at Netflix.
No, sorry. Argument from authority doesn’t mean the original article has a cogent point.
There’s no burden on anyone to refute anything from this piece, as the piece itself has not met any basic requirement of presenting facts or evidence in the first place.
It’s merely a matter of fact to point out this deficiency of the article. The premises of the article could still be accurate (though I think that is fleetingly unlikely), but even if so, this article does not justify any of those claims, so nobody could know one way or the other from this article. Again, this is just a matter of observation of the justifications given.
This author would personally find it more convenient if the skillset of data scientists and data platform engineers coexisted in one person who also happened to have the drive to undertake employment spanning all those skill sets, and wouldn’t become unhappy if the employer did not respect specializations. So this author has decided to read tea leaves out of economic principles and superimpose this wish as if it was justified by some first principles analysis.
In fact, this wishful thinking seems exactly in line with the flawed perspective that executives or director level employees will have. They don’t want to have to care about motivation and intellectual curiosity required to keep certain kinds of knowledge workers happy & productive, and spend lots of time trying to justify how their business units embody corporate platitudes about customer-driven passion. It’s quite easy to see why they would fall victim to this sort of naive wishful thinking. It’s quite similar to CTOs getting suckered by turn-key consulting solutions. It’s not even surprising that VPs & C-suite executives would be very wrong about this type of work.
I like Ray Dalio's principle of "believability."[1] All else being equal, it's reasonable to weight an experienced person's input more than that of a less experienced person.[2]
It depends on what you’re working on. For generic descriptive statistics, then I agree, and also that has nothing whatsoever to do with data science.
If you’re trying to do reverse image search or machine translation or creating custom embeddings unique to your business problem at hand, then deep learning is hands down better.
This bolsters my point as well. If you only hired “full stack” data scientists and you’re trusting them to correctly tell you if / how deep learning is applicable to a new problem, instead of hiring specialists who actually know how to systematically diagnose that situation, you’re setting yourself up to fail. You may already be too biased towards believing simpler things “should” do better, and you’ll take the full stack person’s inability to outperform with deep learning as if it is confirmatory evidence, when really all it is telling you is that you need a specialist.
I somewhat agree with you. Someone who is spending time now studying jQuery and becoming proficient at developing web services would nessesarily not be able to keep up with the pace of deep learning. On the other hand, there are people that had managed to become relatively proficient at developing software a decade ago. And spend last decade at becoming proficient at deep learning.
You need more than a decade to become proficient with deep learning at the level of researchers solving novel business problems.
It takes at least a decade just to study the prerequisite materials in vector calculus, linear algebra, advanced statistics, classifier algorithms, convex and gradient-based optimization, matrix computations and numerical methods, and associated software engineering skills. That’s all just to get to “base camp” of deep learning.
On the flip side, it’s pretty low effort to just use plug-n-play network components from popular libraries and follow a few tutorials or open source projects.
That’s why there’s effectively zero employment demand for the skill of naive keras or pytorch lego building. It’s as easy as it is meaningless.
Given that you’d already have been spending a decade+ of your life on advanced math if you planned to work on deep learning to solve real problems, there’s a huge impedance mismatch with this idea that you’d somehow also magically just be happy ignoring that specialized skill and the time investment sunk into it to then instead be happy writing throw-away little Flask apps or optimizing routine ETL queries.
My assumption was "starting from a post-graduate level in computer science, natural sciences or equivalent". By the way, I don't see how anyone could have more than a decade specifically in deep learning, considering that the field had started at around that time.
On a flip side, TensorFlow 2.0 and AutoML are coming ;). And generic RL agents that do not require reward hacking are also on the horizon. Who cares, if a researcher spend 10000 hours reading articles AND 10000 hours building products, if a more general algorithm obsoletes it all ;)
> “My assumption was "starting from a post-graduate level in computer science, natural sciences or equivalent".”
Yes, same for me. This builds in nearly a decade of preparatory work into the timeline... so it seems we agree.
> “On a flip side, TensorFlow 2.0 and AutoML are coming ;). And generic RL agents that do not require reward hacking are also on the horizon.”
I work professionally in deep learning for image processing. This quote reads like parody to me. I cannot imagine anyone familiar with the realities of AutoML or deep reinforcement learning talking this way. It’s like an excerpt from the script of Silicon Valley.
Yes, I have used AutoKeras in practice, with mixed results. I have also written in-house hyperparameter search tooling to spread parametric architecture search in a distributed training environment with about the same mixed success. I have done this for both large-scale image processing networks and natural language processing networks.
Using AutoML in practice is beyond foolish, given the pricing, except for a really small minority of customers. Let alone that neural architecture search is not a silver bullet and frequently is totally not helpful for model selection (for example, say your trade-off space involves severe penalty on runtime and you have a constraint that your deployed runtime system must be CPU-only.. you may trade performance for the sake of reducing convolutional layers, in a super ad hoc business-driven way that does not translate to any type of objective function for NAS libraries to optimize... one of the most important production systems I currently work on has exactly this type of constraint).
Interesting. I agree, it is not trivial to estimate the runtime of the model on a target device. I wonder how Google does it. They've been boasting about precisely this ability - to optimize for architecture under constraints of precision AND runtime for a target device. And then, claiming that they've been able to get an architecture better than one optimized by a team of engineers over a few years.
It’s all hype coming out of Google. Most of this stuff is meant for foisting overpriced solutions onto unwitting GCP customers who get burnt by vendor lock-in and don’t have enough in-house expertise to vet claims about e.g. overpriced TPUs or overpriced AutoML.
Yes. They were doing 4 years of math intensive undergrad and 5+ years of math intensive post-grad work before there even was an advent of deep learning in 2008-2012.
So they were doing deep learning as an undergrad, huh? Studying tensors for 10 years? Next you’ll be telling me they have 20 years experience coding for CUDA.
No. None of these people have the math background you think, nor do you need it.
I don’t want to belabor this point (but apparently I do, since I’m replying to my own post 2 hours later), but your idea of going back and counting undergraduate work — when it isn’t even related — is simply padding for padding sake. Why stop there? Why not 4 years of high school, too? Aren’t they prerequisites? I mean you can go back all the way to preschool, since counting is a prerequisite to math, and hell, is a whole section of discrete math. But you don’t, because claiming you need 25 years of mathematical training sounds ludicrous.
This reads like a desperate business person wrote it, who wishes that one full-stack set of drives made sense and coexisted in a single person to make that labor cheaper and more commidity, despite the reality that it’s simply not true.
The person who spent the time to master web service frameworks, query languages and product engineering necessarily did not also master professional level knowledge of deep learning or MCMC sampling or natural language processing.
The two types of people need to coexist and work symbiotically, but it’s just asinine wishful thinking to pretend like they are the same person, let alone to write a baseless essay full of assertions that if they aren’t the same person it somehow results in first principles economic inefficiency.