jmount

39 ms·

661.
Using differential privacy to reuse training data(win-vector.com)
2 points by jmount 11y ago|0 comments
662.
Use differential privacy to simulate having more modeling data(win-vector.com)
1 points by jmount 11y ago|0 comments
663.
A Simpler Explanation of Differential Privacy(win-vector.com)
1 points by jmount 11y ago|0 comments
664.
by jmount 11y ago
Moritz Hardt's work is really awesome. Hardt is co-author two of the newer approaches to re-using test data, which I think is going to be a very big thing in machine-learning/data-science.
665.
A deeper theory of testing(win-vector.com)
28 points by jmount 11y ago|5 comments
666.
by jmount 11y ago
"When you fork a repository on GitHub, we create a shallow copy of it. This copy has no objects of its own, but it has access to all the objects of an alternate ..." So does this mean one could attach a GitHub repository by having
667.
How do you know if your model is going to work? P4: Cross-validation techniques(blog.revolutionanalytics.com)
1 points by jmount 11y ago|0 comments
668.
Bitcoin Ponzi scheme operator pleads guilty to securities fraud(theverge.com)
2 points by jmount 11y ago|0 comments
669.
by jmount 11y ago
Way more than "I don't think they disconnected the EKG because I bet we would have seen some electrical noise." The authors wanted to be fooled, else they had no article.
670.
by jmount 11y ago
They didn't. Just because one group of observers didn't find the trick doesn't mean there was no trick.
671.
by jmount 11y ago
Small roundabouts do not work. Here in San Francisco the small ones (out in the burbs) jam up with few cars and are littered with minor car debris.
672.
Is your model going to work? Part 3: Out of sample procedures(win-vector.com)
1 points by jmount 11y ago|0 comments
673.
by jmount 11y ago
It is like they started the essay before they knew they could actually come to a coherent point.
674.
by jmount 11y ago
Looks fun. However, what is REALLY slow in R isn't the interpreter or garbage collector: but some of the object re-allocation patterns you accidentally trigger. Here is my note making some comparisons and recommendations on writing f
675.
Is your model is going to work? Part 2: In-training set measures(win-vector.com)
2 points by jmount 11y ago|0 comments
676.
by jmount 11y ago
Wow, I always wondered how that happened to others. Still no idea what I did.
677.
Bootstrap evaluation of clusters(win-vector.com)
1 points by jmount 11y ago|0 comments
678.
How do you know if your model is going to work? Part 1(win-vector.com)
1 points by jmount 11y ago|0 comments
679.
by jmount 11y ago
I've always said no matter how tough you think you are: never try to do something to somebody that they can't explain to their spouse/partner.
680.
by jmount 11y ago
My demonstration Scala automatic differentiation library: http://www.win-vector.com/blog/2010/06/automatic-differentia...
681.
by jmount 11y ago
A Unix derived operating system with a package manager (BSD Ports, Linux debian/apt or rpm) is way more powerful than one without. Half the time I am installing stuff I don't particularly like or want- but that is necessary to co
682.
by jmount 11y ago
Given spotlight is getting more and more broken, I'd say give "locate" a try.
683.
by jmount 11y ago
Without something like Homebrew you don't get any current Unix tools (OSX shell is way out of date) or anyway to systematically update them. This is a big benefit that far outweighs any perceived benefit dropping bundles into /Ap
684.
by jmount 11y ago
A nice complement to the command line "open" command is adding a shell launcher to OSX finder. I have some instructions how to do this here: http://www.win-vector.com/blog/2012/05/enhance-osx-finder
685.
Systemd Absorbs “su” Command Functionality(tlhp.cf)
43 points by jmount 11y ago|49 comments
686.
by jmount 11y ago
This variant (with the <= comparison) takes c N^2 on lists of length N that a repeated constant ( http://www.win-vector.com/blog/2008/04/sorting-in-anger/ ).
687.
by jmount 11y ago
I think I remember a comedy video where somebody went door to door speaking and messaging in the manner of online advertisements. Probably also perfectly legal (despite any "No Solicitors" signs on front doors) and something tha
688.
by jmount 11y ago
Political dialer is legal, but not legitimate.
689.
by jmount 11y ago
Yep. If your neighbor is willing to enslave their-self to buy a home, then you are going to have to do the same to buy a home on salary (unless you strike it rich).
690.
by jmount 11y ago
Very right. Education, homes- all get more expensive as you make taking loans easier.

More ›