Most research looks more incremental than people think. AlexNet is based on much work by Fukushima, Yann LeCun and others. Likewise, ELMO and BERT are based on 2015 work on pretrained language models:
The Entscheidungsproblem says that proving theorems is at least as hard as the halting problem. This seems irrelevant to the issue of whether human input is essential for finding or proving interesting theorems.
Boy sees a cute girl but can't work up the guts to talk to her.
Boy notices she's taking pictures with Color and writes: 'will you go out with me' on a paper napkin.
Boy takes picture of the napkin with Color.
Girl sees the message but ignores him.
Boy secretly keeps the photos she shared longing for his lost love till he's an old man. Camera pulls back to show his grandkids using Color.
Not anymore. If you receive a targeted attack or a 0-day attack vector through a firefox, pdf or even libpng vulnerability, the only real way you can be safe is to unplug your computer from the internet.
Seems to be an article filled with nothing but hype. It doesn't even define what a random matrix is. I had to first check on Wikipedia to work out that this theory 'probably' has something to do with Wigner's law and symmetric matrices.
Surely one long function is very much harder to test than several shorter ones? So in the context of unit-testing at least, testability is in harmony with Fowler's definition of simplicity. Seems like a nice goal to me.
https://papers.nips.cc/paper/5949-semi-supervised-sequence-l...