Profile cover photo
Profile photo
Gokula Krishnan
103 followers -
Read Problem; Think Hard; Write Solution
Read Problem; Think Hard; Write Solution

103 followers
About
Posts

Post has shared content

Post has shared content
Great news!
An algorithm is not a program, so why describe it as a programming language? 2013 ACM #TuringAward recipient Leslie Lamport is discussing PlusCal -- a toy-like language with more expressive capability than any traditional programming language -- live at the 4th Heidelberg Laureate Forum in Heidelberg, Germany. Stream his lecture live now! http://ow.ly/6gJz304dGjo
Photo
Add a comment...

Post has attachment
Help me win this by registering with this link. THNX
Add a comment...

Post has shared content
Next on my to-watch list. Things get interesting everyday on the interwebz
Deep learning on a Babbage Analytical Engine? Adam P. Goucher got a neural network that can learn to recognize digits working on an emulator of Charles Babbage's Analytical Engine. It fits on 100,000 punched cards, with an extra 40 punched cards for each training image (but only 2,000 images -- fewer than normally used for training neural networks). Even on the emulator, which runs much faster on a modern computer than a real Analytical Engine would, training is very slow. Because of the small number of images, the resulting neural network has overfitting problems.
Add a comment...

Post has shared content
The biggest dangers are from within. Even more reason to not pester sysadmins with requests to fix your personal systems. Keep them happy and they'll keep you productive
The scariest threat is the systems administrator. Sysadmins have godlike access to systems they manage, and are in ideal position to leak, as Edward Snowden has demonstrated. So the NSA will institute a '2-man rule' for all sysadmin positions. It is a concept borrowed from the field of cryptography, where, in effect, two sets of keys are required to unlock a safe.
Add a comment...

Post has attachment
Amazing what Deep Neural Networks can do. DCGAN was also used by the CEO of NVidia in the recent GTC. Interesting work with Chinese alphabets with beautiful visualizations
Add a comment...

Post has shared content
Mr. Robot had an important plot element around using a dropped USB stick
Have you ever found a USB drive left behind in a restaurant or parking lot, or perhaps a  library? Did you pick it up and plug it into your computer in order to find a way to return it? Among the cybersecurity community, there is anecdotal evidence that many people, whether behaving altruistically or due to social engineering, will indeed plug a found USB drive into their computer, exposing themselves (and potentially entire systems) to cyberattack.

But does does this kind of attack actually work or is it merely a myth? To put this attack to the test, researchers from the University of Illinois-Urbana Champaign and the University of Michigan, along with Google anti-abuse & security researcher +Elie Bursztein, dropped nearly 300 USB sticks on the University of Illinois Urbana-Champaign campus and measured who plugged in the drives.

They found that users picked up, plugged in, and clicked on files in 48% of the drives dropped. Furthermore, users did so quickly: the first drive was connected in under six minutes! Head over to Elie's blog, where he summarizes the study, highlights the key findings, looks at what motivates people to plug in USB sticks, and discusses possible mitigations to improve USB security.
Add a comment...

Post has shared content
The more you know!
A brilliant and humorous explanation on the difference between sigterm and sigkill Linux commands.

http://turnoff.us/geek/dont-sigkill/
Photo
Add a comment...

Post has shared content

Post has shared content
Interesting work from Google genomics. The ipython notebooks are worth a read
Google Genomics (http://goo.gl/q4ot5G) has released a dataset that allows you to see how genetic variation is shared among individuals in 26 populations across the world. A Google Dataflow (https://goo.gl/DBRU1W) pipeline computed an analysis on over 5 trillion pairs of variants for each of 31 population groupings of the 2,504 individuals in the 1000 Genomes data set.

To learn more about how Google technologies can be used to to study genomic data, check out the new Linkage Disequilibrium (https://goo.gl/p4Q2bx) data sets at http://goo.gl/A4Od2D and the Jupyter (iPython) notebooks at http://goo.gl/xLh6Sk.
Photo
Add a comment...
Wait while more posts are being loaded