jph00

75 ms·

901.
Practical Deep Learning for Coders(course.fast.ai)
683 points by jph00 10y ago|91 comments
902.
by jph00 10y ago
Just wanted to update this thread to lets folks know that AWS reached out and helped us find a solution for our MOOC students. I've updated the post with this information.
903.
by jph00 10y ago
I wish I'd explained this as clearly as you just did in my post... Thanks!
904.
by jph00 10y ago
It'll be online tomorrow. Keep an eye on http://www.fast.ai for details. There's lots of info there on the original in-person course that it's based on ( https://www.usfca.edu/data-institute/c
905.
by jph00 10y ago
Thanks for looking into it. I just added a couple more concerns to the post: * The totally bizarre responses that requests received. For instance, my co-instructor's request (who in her request included a link to the course and her lin
906.
by jph00 10y ago
Once you know the limit is there, you know to request the increase. For my students, this was less than obvious. Furthermore, decisions as to who were accepted and who rejected were really wacky. For instance, my co-instructor's reques
907.
by jph00 10y ago
> The real problem is that the documentation you've linked to (The EC2 FAQ) actually shows that the p2 instances should have a spot limit of 1, but when I check my account, it's actually 0 for all sizes of p2 instances. Exact
908.
by jph00 10y ago
That's possible, although my understanding is it's more of a capacity planning issue. Indeed, new users can fire up a bunch of much more experience massively multi-core instances and rack up bigger bills than with the GPU instance
909.
Azure and AWS's 'GPU general availability' lies(fast.ai)
41 points by jph00 10y ago|28 comments
910.
by jph00 10y ago
I'm not aware of any systems that I think are much good. It's a big gap.
911.
by jph00 10y ago
I'm really happy to see this. I demonstrated an (in some ways) more advanced version of this a few years ago in my talk on TED.com - see the last few minutes of http://ted.com/talks/jeremy_howard_the_wonderful_and_
912.
Launching fast.ai(fast.ai)
2 points by jph00 10y ago|0 comments
913.
Recurrent/Recursive Nets (Video Tutorial in Python)(youtube.com)
2 points by jph00 10y ago|0 comments
914.
by jph00 11y ago
There's a few questions that I had that I'm sure others would have too, but I couldn't find any answers on your site: * What's the maximum upload size? * If I want to change just one file later, do I have to upload my wh
915.
by jph00 11y ago
If anyone wants to get this running on Windows, I've made a start. In cpp\thirdparty, you can run the .sh scripts with cygwin, but in build_thirdparty you'll need to add to the appropriate section: elif [[ "$OSTYPE&quot
916.
by jph00 11y ago
I think "machine learning" is more accurate and useful, personally.
917.
by jph00 11y ago
I don't think we have any reason to believe that strong AI is impossible - in fact, I can't see how it could be impossible (since we have an existence proof, in ourselves, that it's possible!) But I don't see any way to
918.
by jph00 11y ago
The difference with deep neural networks is that they can do so much of their own feature generation. It's very different to bayesian networks or SVMs. Because they can create their own features (to a large extent) they can handle unst
919.
by jph00 11y ago
Sure. Deep learning is a tool that can impact a large percentage of processes that we use to create and provide products and services. Many of these processes involve human steps because humans are uniquely able to provide the perception, l
920.
by jph00 11y ago
I'm Jeremy Howard, one of the people quoted in the article (and the one in the crazy photo!) AMA about applications of deep learning. My views from 18 months ago are in this talk on TED.com: http://www.ted.com/talks&#x2
921.
by jph00 12y ago
Company: Enlitic http://www.enlitic.com/ Location: San Francisco REMOTE is fine if you're happy to visit regularly If you have world-class machine learning, math, programming, stats, or physics skills, and want to use
922.
Deep learning in diagnostic healthcare: The future?(idgconnect.com)
7 points by jph00 12y ago|1 comments
923.
by jph00 12y ago
That seems like a stretch - according to Wikipedia : "The magazine is published by Technology Review, Inc, an independent media company owned by MIT. MIT's website lists it as an MIT publication, and the MIT News Office states tha
924.
by jph00 12y ago
Oh I should also mention - if you're interested in playing with cool hardware and data, we're a great place to come. We use a cluster of hundreds of Tesla K80 GPUs, and have many millions of medical images - some single images are
925.
by jph00 12y ago
Company: Enlitic http://www.enlitic.com/ Location: San Francisco REMOTE is fine if you're happy to visit regularly :) If you have world-class machine learning, math, programming, stats, or physics skills, and want to u
926.
by jph00 12y ago
Wonderful! Despite being the founder of FastMail, I had to stop using it after we sold FastMail and I moved to Kaggle, due to lack of calendar support and contacts sync. I even wrote my own FastMail contacts sync service and kept supporting
927.
The wonderful and terrifying implications of computers that can learn (TED)(ted.com)
4 points by jph00 12y ago|0 comments
928.
by jph00 12y ago
I would start with the Intro to Machine Learning Coursera course, and then the Intro to Neural Networks course. That will get you to a point that you can understand current papers. The demo I showed actually showed that it's possible t
929.
by jph00 12y ago
I'm the speaker from this TEDx talk (and Enlitic CEO / Kaggle Past President) - happy to answer any questions or respond to any comments about it here. Sorry for some rather extreme simplifications in the talk... I only had 18 min
930.
The wonderful and terrifying implications of computers that can learn(youtube.com)
9 points by jph00 12y ago|4 comments

More ›