jm + obfuscation   5

Google release an open-source differential-privacy lib
Differentially-private data analysis is a principled approach that enables organizations to learn from the majority of their data while simultaneously ensuring that those results do not allow any individual's data to be distinguished or re-identified. This type of analysis can be implemented in a wide variety of ways and for many different purposes. For example, if you are a health researcher, you may want to compare the average amount of time patients remain admitted across various hospitals in order to determine if there are differences in care. Differential privacy is a high-assurance, analytic means of ensuring that use cases like this are addressed in a privacy-preserving manner.

Currently, we provide algorithms to compute the following:

Count
Sum
Mean
Variance
Standard deviation
Order statistics (including min, max, and median)
analytics  google  ml  privacy  differential-privacy  aggregation  statistics  obfuscation  approximation  algorithms 
9 days ago by jm
Minecraft now publishing deobfuscation maps
About time too.
In an effort to help make modding the game easier, we have decided to publish our game obfuscation maps with all future releases of the game, starting today. This means that anyone who is interested may deobfuscate the game and find their way around the code without needing to spend a few months figuring out what’s what. It is our hope that mod authors and mod framework authors use these files to augment their updating processes that they have today. These mappings will always be available, instantly and immediately as part of every newly released version. This does not, however, change the existing restrictions on what you may or may not do with our game code or assets. The links to the obfuscation mappings are included as part of the version manifest json, and may be automatically pulled for any given version.
minecraft  obfuscation  microsoft  mods  modding  community  coding  games 
13 days ago by jm
A looming breakthrough in indistinguishability obfuscation
'The team’s obfuscator works by transforming a computer program into what Sahai calls a “multilinear jigsaw puzzle.” Each piece of the program gets obfuscated by mixing in random elements that are carefully chosen so that if you run the garbled program in the intended way, the randomness cancels out and the pieces fit together to compute the correct output. But if you try to do anything else with the program, the randomness makes each individual puzzle piece look meaningless. This obfuscation scheme is unbreakable, the team showed, provided that a certain newfangled problem about lattices is as hard to solve as the team thinks it is. Time will tell if this assumption is warranted, but the scheme has already resisted several attempts to crack it, and Sahai, Barak and Garg, together with Yael Tauman Kalai of Microsoft Research New England and Omer Paneth of Boston University, have proved that the most natural types of attacks on the system are guaranteed to fail. And the hard lattice problem, though new, is closely related to a family of hard problems that have stood up to testing and are used in practical encryption schemes.'

(via Tony Finch)
obfuscation  cryptography  via:fanf  security  hard-lattice-problem  crypto  science 
february 2014 by jm
Obfuscatory pie-chart from Garda penalty-points corruption report
"Twitter / gavinsblog: For sake of clarity here is helpful pie chart of the 95.4% of fixed charge notices not terminated #missingthepoint"

Paging Edward Tufte: classic example of an obfuscatory pie-chart, diagramming the wrong thing misleadingly. By presenting it like this, it appears that the 95.4% of cases where fixed charge notices were issued by the guards are relevant to the discussion of the other classes; in reality, that means that 4.6% of cases, 37,000 cases, were terminated, some for good reasons, others for not, and it's the difference between those two classes that are relevant.

In my opinion, 2 separate pie charts would be better; one to show the dismissed-versus-undismissed count (which IMO could have been omitted entirely), and one to show the good-vs-not-so-good termination reason counts (which is the meat of the issue).
dataviz  visualisation  data  obfuscation  gardai  police  corruption  penalty-points 
may 2013 by jm
Stuxnet is embarrassing, not amazing « root labs rdist
interesting post from Nate Lawson -- he suggests that Stuxnet could have been much better in payload obfuscation, had the authors studied the state of the art in malware implementation.  I'm not convinced, however; as Halvar Flake suggests, KISS applies
kiss  stuxnet  security  malware  obfuscation  siemens  from delicious
january 2011 by jm

Copy this bookmark:



description:


tags: