Profile

Cover photo
David Lehavi
Works at HP Labs
Attended Hebrew University Of Jerusalem.
108 followers|14,341 views
AboutPostsPhotosVideos

Stream

David Lehavi

Shared publicly  - 
 
 
For quite sometime we are trying to figure out what's the best way to document code which calculate complicated mathematics.
Trying to explain the formula with words is hard, external documentation is easy to read but hard to keep updated (and is out of context).
Yesterday I saw a post on a chrome extension which converts latex formulas in FB posts, this is not very useful for me  but then I thought that the some on bitbuckt (where I do most of my code reviews these days) would be great.
So I sat down to write my first chrome extension, and here it is.

Hope you'll find it useful as well :)
1
Add a comment...
 
quoting from http://www.yosefk.com/blog/ latest post (which is always worth reading - not only for humour): "In this sense, C++ is fine and a worthy achievement of a lifetime. Especially in a world where Putin is a candidate for a Nobel Peace prize and Obama already got one."
3
Elazar Leibovich's profile photoDavid Lehavi's profile photo
3 comments
 
Ada actually looked pretty good back then, and is successfully used in practice, but since there's almost zero support for it in the industry, it's not enough unfortunately.[0]

And heck, your kernel consumes C strings, which is probably the worst possible representation, so you're pretty much trapped.

[0] http://gcn.com/Articles/2008/04/11/The-return-of-Ada.aspx?Page=1
Add a comment...

David Lehavi

Shared publicly  - 
2
3
David Lehavi's profile photoArik Baratz's profile photoEyal Rozenman's profile photoSagi. Schein's profile photo
14 comments
 
+Sagi. Schein Ye, I bet this would happen....
Add a comment...

David Lehavi

Shared publicly  - 
2
1
Max Lieblich's profile photo
Add a comment...

David Lehavi

Shared publicly  - 
6
1
Rani Amit Bar-On's profile photo
Add a comment...
Have him in circles
108 people
Nimrod Lehavi's profile photo
Karni Gilon's profile photo
Omer Tamuz's profile photo
Haggai Scolnicov's profile photo
Carl Staelin's profile photo
Eyal Rozenman's profile photo

David Lehavi

Shared publicly  - 
 
The internet is so full of Pete Seeger today, so I feel I have to balance it: 
The Folk Song Army by Tom Lehrer
2
1
Noam Nisan's profile photo
Add a comment...

David Lehavi

Shared publicly  - 
 
Back to my recuring column: Complaining about education. Since we are now back in Israel, this time it's going to be Israeli education; the  kids are now in "HaReali" (non Israeli readers: posh priavte school with good STEM corriculum).

The good: I did not follow closely all the STEM classes the kids take, but the ones I did follow (math, CS, physics in High, honors math in middle, "Science" in elmentary) are between OK and good.

The bad and the ugly: We went today to see the so called "learning products" in middle school geography. The current fad around here in social studies is "project oriented learning". We've experienced this before both in Israel and in the US, and were never overly impressed: Seem very good for self learners - for whom almost anything is very good - and not that good for the rest (that being said, it is very much in the tradition of good US liberal arts colleges - which are rather good - so maybe by the end of high school, kids actually learn in this method when properly applied). Anyway, our middle schooler and the teachers stressed how important it is that we'll see the "learning products" on which the kids worked the entire semester. As it turn out, the projects (done in groups of four) was to film a sketch comparing a first world country and a third world country. All the sketches exhibited middle school humor at it's best (or worst - depending where you taste swings); and all exhibited complete igonarnce. Our kid's sketch was on Brussles, Bucharest, and around the claimed "fact" that 30% of Belgium export is chocolate (I seriously doubt that). As it turned out when we came home, he does not know that Brussles is in Belguim, and when asked to locate Belguim on the world map he put it in Russia. 
3
1
Yoad Lustig's profile photoElazar Leibovich's profile photoDavid Lehavi's profile photo
2 comments
 
+Elazar Leibovich That's the symptom; not the problem. I see two problems:
a) Being blinded by technology (we were in a very strong US school district last year, schools are deeply immersed in technology there, but nobody is blinded by it).
b) Strong preference of form over content.
Add a comment...

David Lehavi

Shared publicly  - 
 
http://www.theroboticschallenge.org/

is over, with very impressive results.
1
2
Noam Nisan's profile photoDimitris Vardoulakis's profile photo
Add a comment...

David Lehavi

Shared publicly  - 
 
 
So freaking cool.  
Called the Cubli, it measures nearly 6 inches on each side and can walk around by continually flipping itself over. Perhaps more impressively, it can also balance on any of its sides or even just a single corner
2
3
Omer Woloch's profile photoRani Amit Bar-On's profile photo
Add a comment...

David Lehavi

Shared publicly  - 
2
1
Elazar Leibovich's profile photoDavid Lehavi's profile photoEyal Rozenman's profile photoEyal Fink's profile photo
16 comments
 
Well, it's obviously fiction, but very much based on reality - That's what CSI is about (not the TV series, the real thing): Even in the most optimistic case, the investigators have all the prints they found near the crime scene, as well as all the faces from CCTV's in the nearby streets. This data is then cross-reffed to the  bioDB, and the cross reference sits in the police station for years. In the meanwhile they use this data to obtain more suspects, harvest more prints and pictures, and so on and so forth. Now multiply this by the amount of open investigations. Your DB leaked just because you can automatically make mass queries - ain't science something ?
Note that we played this game And without letting a single image out of the DB, when the legislation (please read it - I gave the link somewhere in this thread) alows tons of people to pull the bio data out - so why wouldn't they ?

Phrased in another way (and directly answering your claim about using the DB w/o exposing it): there are two angles here: the Human angle, and the "provable security" angle. I just showed you that the DB - even under a milder legislation than the suggested one - is provably insecure. That is a reason +Eyal Fink started the conversation with "I know it's going to leak"

Counter scoop: the fact that you can obtain data for a certain price about almost anyone is feature not a bug. The fact that this price is non trivial (to a back of the envelope estimate, don't just say "it's not hi" - you are simply wrong here) is what's currently stopping us from being under constant surveillance (which will include our medical and financial data, things I - and most people - like to keep to themself).

The database the military has is much better legislated (to be used only to identify your body, this was violated only <10 times ever AFAIK), and it is indexed by name.
Please don't bring the current US as a model for where you want to go. I'm sure you can read the news about the current mischiefs of the NSA and the drone command.

I am not a security expert either, but a) I know enough - some of what I know I told you, and b) most of the security people I know are either terrified, or thinking how to use the situation to their advantage (money, consulting, getting grants - Adi Shamir being a notable example here). If you think I'm wrong, a good counter argument is not "there are procedures", but "here are the procedures, and this is why they will work under any given attack".
Add a comment...
People
Have him in circles
108 people
Nimrod Lehavi's profile photo
Karni Gilon's profile photo
Omer Tamuz's profile photo
Haggai Scolnicov's profile photo
Carl Staelin's profile photo
Eyal Rozenman's profile photo
Education
  • Hebrew University Of Jerusalem.
    B.Sc., M.Sc., Ph.D. in Math, 1991 - 2002
Work
Occupation
Senior research scientist at HP labs Israel
Employment
  • HP Labs
    Senior research scientist, 2010 - present
  • Correlix
    Senior algorithm developer, 2007 - 2010
  • University Of Michigan math dept.
    Visiting scholar, 2006 - 2007
  • Princeton University math dept.
    visiting assitant professor, 2005 - 2006
  • Ohio State University math dept.
    Zassenhaus assistant professor, 2002 - 2005
  • Compugen
    algorithms developer, 1998 - 2001
Basic Information
Gender
Male