Relates surprisingly well to the next thing in my feed.https://twitter.com/LizardLizardx/status/951999465828864000 …
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
Relates surprisingly well to the next thing in my feed.https://twitter.com/LizardLizardx/status/951999465828864000 …
Deep learning propaganda deserves all the skepticism that religion does for the same reasons, most of it is bunk that the public accepts based upon flimsy understanding.
... the main argument is: it cannot abstract and generalize, even in face of simple algebraic relations instances. Isn't this the same fundamental tenet of Minsky's criticism of Rosemblatt's Perceptrons?
This brings to mind something a wise professor told me: linguists need to study animal behavior: communication is in context of sex, dominance, hunger, etc. Deep learning researchers have the same myopia. Perhaps pair DeepMind with evolving Game of Life, prisoner’s dilemma.
I've read these principled arguments against neural net models since 1987 when I first encountered the PDP books. Arguments about symbol systems have always struck me as persuasive. Meanwhile, neural net (aka "deep learning") models just keep doing what people say they cant do.
Arguments in principle will pale in comparison to the practical achievements of the deep learning models. Especially in domain of language, where symbol systems presumably should have have triumphed, all the major achievements of the past 30 years have been from deep learning.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.