jph00

65 ms·

811.
by jph00 9y ago
I agree with everything you said. However, I don't agree it's what should be hogging the money and interest right now. The impact of AI on employment (and therefore societal stability) is a much more pressing problem. As is the mi
812.
by jph00 9y ago
It's because we care about our mission and want to support it and promote it. We're entirely self funded from our own personal finances, so trying to suggest we have some other purpose to our work is a little odd... We want to bri
813.
by jph00 9y ago
Yes exactly. Fewer people can add economic value at that point, and unless there is a negative income tax or basic guaranteed income, inequality spirals out of control, leading to even more of the anger-driven policy we're starting to
814.
by jph00 9y ago
Oh please. The issue isn't what the end result could look like, but whether we'll actually get there if basic societal needs aren't met first. Prioritize things which are needed now and in the near future.
815.
by jph00 9y ago
That's not how things have generally worked out in the world. Billions live in abject poverty even although the world has plenty enough to feed them and provide shelter. The issue is income distribution - and just making more money doe
816.
by jph00 9y ago
Absolutely! We'd like to see investment in both. Currently we're just not seeing the investment we think there ought to be in the areas Rachel mentioned, or in application areas around food production, education, medicine, and so
817.
by jph00 9y ago
We also do research and try to move the state of the art. Although we share our results and methods largely through courses rather than academic papers. Our plan is to teach more material each year, in less time, with less prerequisites, by
818.
by jph00 9y ago
Planning to release it on Sunday night :)
819.
by jph00 9y ago
Perhaps, but the jury is still very much out. The vast majority of RL applications are game playing. Very few examples of valuable applications to society or the economy. There's also plenty of evidence already that RL isn't reall
820.
by jph00 9y ago
We don't make specific time predictions because it's just not possible. But we can make relative predictions. However long it takes to get to AGI, when we're (say) half way there, the amount of societal upheaval will be enorm
821.
by jph00 9y ago
Can you be more specific? China is a big place and there's a lot happening there! One non-profit in China for instance has kindly translated the entirety of part 1 into Chinese and provided a discussion group for Chinese students.
822.
by jph00 9y ago
Rachel and I (fast.ai co-founders) will be keeping an eye on this thread, so if you have any questions or thoughts, let us know!
823.
by jph00 9y ago
Whilst what you say is accurate, it has nothing to do with my comment which you are replying to, which has specifically about the shape of the loss functions in deep learning, and recent research on how it impacts optimization of these func
824.
by jph00 9y ago
Exactly. Saddle points are like a magnet for them!
825.
by jph00 9y ago
Yes it doesn't directly cover ML much. It does assume some basic linear algebra, but provides links for where to learn that quickly.
826.
by jph00 9y ago
Sorry - I meant: videos 9 & 10.
827.
by jph00 9y ago
Krylov spaces often come up in derivations for optimization methods in DL, and have been studied as an explicit technique on their own. Eg see https://arxiv.org/abs/1111.4259 However more recent theoretical and empiric
828.
by jph00 9y ago
It's a very helpful exercise, and something we do in lesson 9 of this course.
829.
by jph00 9y ago
Absolutely right. There's very little linear algebra required in deep learning. This Computational Linear Algebra course mainly covers decompositions, which aren't used much in deep learning.
830.
by jph00 9y ago
Specifically, it's sections 9 and 10 here: https://github.com/fastai/numerical-linear-algebra/blob/mast...
831.
by jph00 9y ago
Krylov subspace methods are heavily covered in the course. Trevathan is the main text we recommend in the course.
832.
by jph00 9y ago
Yes, the memory hierarchy is discussed in the course, and algorithms are shown that are designed to take account of this. Older algorithms tended to only consider computation time, not memory access time.
833.
by jph00 9y ago
Gradient descent is covered. We even show a pytorch implementation :)
834.
How to piss off the AI talent you're trying to recruit(topbots.com)
1 points by jph00 9y ago|0 comments
835.
by jph00 9y ago
Oh also I did say that CNNs often beat RNNs, even for time series, language, and audio. So I generally try them first. You can always stick an RNN on top of the CNN and get the best of both worlds.
836.
by jph00 9y ago
I hope that wasn't quite what I said :) I said, IIRC, that they're very similar in terms of results and GRUs are a little simpler. I've seen some papers show better results for GRU, and visa versa.
837.
by jph00 9y ago
FYI the OP also created a really nice way to create persistent spot instances on AWS - see approach 2 here: http://wiki.fast.ai/index.php/AWS_Spot_instances
838.
by jph00 9y ago
Just to clarify - the actual url is course.fast.ai :)
839.
by jph00 9y ago
To claim that those techniques are even in the same ballpark as deep learning based style transfer is just silly. They are extremely limited in their application, as you can clearly see by looking at the examples. And the entire approaches
840.
by jph00 9y ago
Because in high dimensional space there are partial saddle points, not local minima. Intuitively obvious, and shown theoretically in the last 12 months.

More ›