Monday, November 08, 2010

WIRED: November Jargon Watch & Forensics?

One of my NASA buddies (hi, Lisa!) dropped by last week for coffee and to catch up on the world of information management. When I introduced her to one of the PhD candidates in our lab, Brad Wardman, she dropped a stray comment "Oh, have you been following the MIT Probability Chip? It seems relevant to what you are doing..."

I haven't asked, but she may have heard about the chip in this month's WIRED magazine. Although I browse lots of magazines, I have to confess the only ones I read cover to cover every month are WIRED and Analog.

This month, three of the four Jargon Watch terms had potential Forensic Applications, so I'm doing a bit of an odd column here and talking about them in more detail. (since they aren't actually online yet, I'm not going to "quote" them. Here's a link to the last available (October) Jargon Watch which has nothing to do with this article other than to give a shout out to Jonathan Keats!) Its one of my favorite columns in WIRED each month.

For those who don't read WIRED, Jargon Watch looks for new science and technology words or phrases that are beginning to be used more broadly in the media. The three from the November WIRED that I want to dig a bit deeper on were Bacterial Fingerprints, Probability Chip, and Cybercase. I first looked at the Probability Chip because of the hint Lisa dropped, but had so much fun doing it, I decided to dig deeper on the other two as well.

Bacterial Fingerprinting

Bacterial Fingerprinting is the idea that your fingertips may have a unique bacterial colony that could be retrieved from items you touch, such as your keyboard or mouse, and used to identify you. It started inching into the public consciousness with a CBS News story back in March that covered research at the University of Colorado by Dr. Christian Lauber. According to the story, the researchers gathered swabs from the keyboards and mice of three people and compared the bacteria found there to that found on the hands of 270 random people. 87 percent of the bacteria is unlike anyone else's, according to the story. Science Daily had more facts. The researchers, who are actually at University of Colorado at Boulder had their study published March 15th in the Proceedings of the National Academy of Sciences. Noah Fierer was the lead author, with Christian Lauber and Nick Zhou. They linked together chemistry and biochemistry departments for the study. In a second phase of the study, they sampled nine computer mouses that had not been used for more than 12 hours, and were able to find their owners when mixed with the same group of 270 random people.
Fierer's Lab seems to be a publishing MONSTER on this and related topics. Here is the story "Forensic identification using skin bacterial communities", which is the BEST source for all of the above.

According to Science Daily, in an earlier study in 2008, 4,700 different bacteria species were found on 102 human hands, with only five species being shared by all participants in the study. That study was actually The influence of sex, handedness, and washing on the diversity of hand surface bacteria, which planted the seeds that your personal bacterial colony may be of forensic interest.

The Probability Chip

The Probability Chip is a new type of microprocessor that claims to not use the traditional 0 and 1 that has been computing's mainstay since the days of the vacuum tube, but rather a new type of logic gate that calculates the probability that a value is 0 or 1. The story was covered by ZDNet in this recent story "Start-up sets sights on probability chip". For Jargon Watch to be interested, things have to have moved out of the academic community and into "the public eye." The story feature Lyris Semiconductor, an MIT startup.

On the company's website, they claim that with this new approach "many applications that today require a thousand conventional processors will soon run in just one Lyric processor, providing 1,000x efficiencies in cost, power, and size."

They go into more details, showing a demo that uses their new programming language that will take advantage of this new form of chip, called "PSBL" or "Probability Synthesis for Bayesian Logic".

So where is the Forensics link? They also have a demo about clustering where they have various typists type on a keyboard and log the timing and sequence of their keystrokes to create a forensic signature of their typing style. They claim "probability processing" would be especially strong at this type of calculation, and then they go on to imagine spam filtering as well:

"...It can tell which text was entered by which person. In the real world, this could help identify unauthorized access to a computer merely by observing the rhythms and typing habits of the designated user, and determining when someone else is accessing their computer.

This same class of computations can be used to cluster data for applications ranging from network security, to spam filtering, to enterprise search."


Cybercasing is the idea that when one shares pictures or information online, it could be used by a potential thief to determine the geographic location of the item they would like to steal. Additional information about that location could help them "case the joint" and determine ideal times and methods of gaining access. This was described in The Atlantic's story "How Tech Savvy Thieves Could Cybercase Your House" which quoted a paper published in August at the Fifth USENIX Workshop on Hot Topics in Security (HotSec 10) by Friedland and Sommer of Berkeley's International Computer Science Institute (ICSI): "Cybercasing the Joint: On the Privacy Implications of Geotagging". You can read their full paper from the author's website, or see an Abstract or the Slides from the HotSec conference.

Gerald Friedland went on to create the website talking about some of the geotagging things we do (and possibly shouldn't) including taking GPS-labelled photos (Geotagged) with our iPhones, and playing "Foursquare", a game where people "check in" to let the world know their exact geographic location at all times.


So, go subscribe to WIRED magazine!

No comments:

Post a Comment

Trying a new setting. After turning on comments, I got about 20-30 comments per day that were all link spam. Sorry to require login, but the spam was too much.