3 ms·
No need to be super pessimistic, but there is reason to temper your excitement - this is quite preliminary research (little results to indicate much benefit to
by andreyk 9y ago
No need to be super pessimistic, but there is reason to temper your excitement - this is quite preliminary research (little results to indicate much benefit to this - "But the results show that even on the ImageNet task, the presence of such blocks does not detract from performance, and may even slightly improve it."), and is still entirely supervised. Something like this able to learn in a semi-supervised fashion from images and text could really seem revolutionary.
- sdenton4 9y agoThere's a pretty strong literature around retraining just the top couple layers of a deep net to Target a slightly different objective. The interesting possiblity opened here is training new feature processing frontends to work with an established 'conceptual' backend.
- innagadadavida 9y agoIf AI really really, worked, Google would’ve called the team: “Google Brainless”. Hang on tight we are getting there.
- tripzilch 9y agoTo be fair, their ImageNet results were 86% versus 95% in state of the art. In ML, those last few percentages, the closer you get to 100% become exceedingly harder to beat.