jph00

144 ms·

931.
by jph00 12y ago
Could you describe what facilities of R help with metaprogramming and make it good for designing DSLs?
932.
by jph00 12y ago
I wasn't the person that made the claim - but since I feel the same way (well, maybe not quite as strongly!)... I find I often lose track of earlier messages in a thread that I haven't yet responded to, because I reply to some oth
933.
by jph00 12y ago
...or at least a description on your actual web page? All I see there is a change log and example.
934.
by jph00 12y ago
Well said. That's pretty much what I'm trying to do with my new startup, Enlitic: http://www.technologyreview.com/news/530261/a-startup-hopes-... Furthermore, the things the author of the NYT piece compl
935.
With Enlitic, a veteran data scientist fights disease using deep learning(gigaom.com)
2 points by jph00 12y ago|0 comments
936.
Data Scientist on a Quest to Turn Computers into Doctors(wired.com)
7 points by jph00 12y ago|0 comments
937.
Computer Eyesight Gets a Lot More Accurate(bits.blogs.nytimes.com)
2 points by jph00 12y ago|0 comments
938.
The future of big data(mobile.nytimes.com)
4 points by jph00 13y ago|0 comments
939.
I don't care how big your data is(forumblog.org)
1 points by jph00 13y ago|0 comments
940.
Is Big Data an Economic Big Dud?(mobile.nytimes.com)
1 points by jph00 13y ago|0 comments
941.
An Inventor Wants One Less Wire to Worry About(mobile.nytimes.com)
1 points by jph00 13y ago|0 comments
942.
We’re on the cusp of deep learning for the masses. You can thank Google later(gigaom.com)
5 points by jph00 13y ago|0 comments
943.
by jph00 13y ago
I just gave an invited talk at KDD about deep learning in which I covered this algorithm, so it's great to see this code appear now. For anyone interested in text analysis: PLEASE study and use this code and the referenced papers. It&#
944.
by jph00 13y ago
Handwritten digits actually is a pretty good domain for deep nets, and the poor performance achieved in the article's case is due to the implementation (it needs deeper net, convolutional layer, etc). In that case much better (99%+) re
945.
What Kind Of Data Scientist Are You? (A survey of 250 data scientists)(fastcolabs.com)
1 points by jph00 13y ago|0 comments
946.
Obama issues executive order making open data the new default(e-pluribusunum.com)
1 points by jph00 13y ago|0 comments
947.
by jph00 14y ago
A lot of commenters here seem to have missed this: "The really disruptive thing about Kaggle, though, comes through the company's new service, Kaggle Connect. Here, Kaggle acts as a match-maker, where customers with a specific problem can h
948.
by jph00 14y ago
Oh that's annoying. When I did the webinar I was sharing my screen and showing both slides and scribbles in onenote. Is there no button to view that in the recording? Sorry I can't check myself now, I'm on my phone.
949.
by jph00 14y ago
There's also the random knockout of neurons, as mentioned in the webinar.
950.
by jph00 14y ago
DNNs can be thought of a stacked Restricted Boltzmann Machines. Their structure and training is very different to traditional MLPs. They derive in some ways from convolutional neural nets. I describe some of the key differences between DNNs
951.
by jph00 14y ago
I believe the talk about the talk only covers 3 min 20 s.
952.
by jph00 14y ago
I don't think so. I even had to register to view - and it's my talk! I haven't received any marketing stuff from Cloudera or O'Reilly however. Honestly, I doubt those companies would do anything questionable with registrations.
953.
by jph00 14y ago
I created this talk for the Enterprise Big Data track of O'Reilly's Strata conference - so it's not a technical description of how deep learning works. Rather, it's an attempt to show why it's important, and how it fits into current data sc
954.
Deep Learning - The Biggest Data Science Breakthrough of the Decade(oreillynet.com)
127 points by jph00 14y ago|73 comments
955.
by jph00 14y ago
I'm the President and Chief Scientist of Kaggle, which ran the drug discovery project mentioned in the article. As it happens, I did my Strata talk on Tuesday about just this topic. I will be repeating the talk in webcast form (for free) in
956.
GSK announces support for AllTrials campaign for clinical data transparency(gsk.com)
5 points by jph00 14y ago|0 comments
957.
by jph00 14y ago
Huge. When I'm sleepy my error rate can go up by as much as 300%. It's terrifying to think of what that means in terms of how it must be impacting by work when I see that happen. I often study in the morning before work, and if I can see I'
958.
by jph00 14y ago
I only have quantified outcomes for studying flash cards, since that's a fixed measured thing that I do for 70 minutes per day. However I now do nearly all my work at the tread desk, and I've bought a few for some of my staff who also wante
959.
by jph00 14y ago
I tracked my learning performance over two years under a range of conditions, including trying a treadmill desk. I found at 1.2mph my error rate (compared to sitting) was 50% less, and I could comfortably study for 2x as long. The trick is
960.
by jph00 14y ago
The original Chinese actually uses English for the model names in this article, so we have nothing more to go on...

More ›