i'd love to be proved wrong but it feels less and less likely that there will be any research around DL on small compute/less data - Google, FB et al are not interested - they want everybody to use their cloud and pre-trained models discuss
-
Show this thread
-
Replying to @glagolista @asthesparksfly
Really good ideas - ideas as good as backprop, GANs, conv nets, and so on, in their time - seem likely to be much _harder_ to have with big compute/big data. Longer iteration time, more tempting to lean on your resources, more specialization in those resources.
1 reply 0 retweets 7 likes -
It feels a bit like the argument that big established companies will always beat small startup companies. All other things equal, but the trick for the small companies is to make sure all other things aren't equal. And they have a lot of latitude to do this.
1 reply 0 retweets 2 likes -
Replying to @michael_nielsen @asthesparksfly
i'm not so much talking about the company size but more so about the research - these days AI/DL research is pretty much driven by a few hi-tech companies
1 reply 0 retweets 1 like
I was making an analogy in my second tweet, not a direct comparison.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.