You can add location information to your Tweets, such as your city or precise location, from the web and via third-party applications. You always have the option to delete your Tweet location history. Learn more
Very really revealing thread, as DL field try immunizes itself from candid talk about it limits -- by saying both that (A) we already know the limits (@ylecun) but also that (B) that there are no known limits (@roydanroy, @earnmyturns).
These can't both be true.https://twitter.com/GaryMarcus/status/1173377561897816065 …
Ps hats off to @ylecun for being the most candid and actually answering (rather than evading) my original question, by listing multiple limits. I largely agree with his reading of the literature and inserted a few comments embedded in the thread.
They can both be true. There former was talking about practical limitations, whereas the latter were discussing fundamental limitations. Saying what can't be done right now is much easier than saying what can't be done (even within a certain class of methods).
so what are the practical limitations you would acknowledge?
Yes they can. Yann is speaking informally and I'm speaking formally. And I think you know that. There are few theorems expressing limitations and when they do, they are quite cryptic.
deep learning folks often hide behind that when they speak in public, and the public becomes misled.
Your response is revealing of your bad faith. As I said, every single one of my talks of the last 4+ years have focused on the limitations of supervised & reinforcement learning (deep or not), and on ways to lift them (e.g. with self-supervised learning)....
i gave you full credit for a candid answer multiple times in this thread; your reply here is bizarre.
DL is not the right paradigm for rare conditions which collectively affect up to 10 % of population. The models have an insatiable appetite for data risking further entrenchment of health inequities - the herd follows big datasets at the expense of all else
I don’t know what “DL” here refers too. Can you be more precise?
It's possible people are interpreting it differently. There are endless limits to what we can do in practice now. There are also limits to what vanilla end-to-end FA can do. At the same time, it seems likely we'll continue to find new tasks we can hit with the FA big data stick.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.
