Research shows eye-reflections in photos could be used to identify criminals

Did you ever watch CSI's tech boffins zoom into a photo, grab the reflection in someone's eye, and through processing magic generate a full facial image of the wanted criminal? If, like us, you rolled your eyes at the program's "Hollywood" technology, then maybe you did so too soon (again). Researchers at the University of Glasgow, Scotland, have effectively done just that. Okay, so the team was using near perfect conditions: a 39-megapixel Hasselblad H2D camera, shooting a subject from one meter away, in perfect lighting, but the results proved the idea is well within the realms of science fact. When looking at a close up of a subject's eye, the team was able to grab an image between 27 to 36 pixels wide by 42 to 56 pixels high showing the face of one of the "bystanders" they had organised to be in the subject's view. The reflected images proved good enough that when volunteers were asked to match the small image with that of the actual bystander, or someone of similar appearance, results came in well above chance averaging between 70- and 84 percent. As well as assisting in criminal investigations, the researchers claim the technique could also yield 3D environments, when images can be taken from both eyes. Of course, real-world application would require similarly optimal conditions, but as technology advances, the researches claim this could be an ever increasing scenario.

0 Comments Share

Via: Sky News

Source: Plos One

Tags: eye reflections, photo, research, University of Glasgow Next: Roku player software cracked open temporarily, root now to run XBMC later .fyre .fyre-comment-divider

View the Original article

Stanford Researcher Proves NSA Can Probably Identify Individuals From Phone Records

The National Security Agency likes to claim that intelligence officers are only collecting the phone records of millions of Americans, safely omitting their actual names from analysis. But a Stanford researcher, Jonathan Mayer, found that he and his co-author could easily match so-called “meta-data” to individual names with little more than a Google search.

“If a few academic researchers can get this far this quickly, it’s difficult to believe the NSA would have any trouble identifying the overwhelming majority of American phone numbers,” they wrote.

Using a crowdsourced public database of voluntarily submitted phone records, MetaPhone…

We randomly sampled 5,000 numbers from our crowdsourced MetaPhone dataset and queried the Yelp, Google Places, and Facebook directories. With little marginal effort and just those three sources—all free and public—we matched 1,356 (27.1%) of the numbers. Specifically, there were 378 hits (7.6%) on Yelp, 684 (13.7%) on Google Places, and 618 (12.3%) on Facebook.

What about if an organization were willing to put in some manpower?

To conservatively approximate human analysis, we randomly sampled 100 numbers from our dataset, then ran Google searches on each. In under an hour, we were able to associate an individual or a business with 60 of the 100 numbers. When we added in our three initial sources, we were up to 73.

The science of identifying people from supposedly anonymous databases has become a game for academics. Last year, a group of researchers proved that they could identify individuals from a DNA database of their relatives and public demographic information.

On the more invasive side, other researchers could estimate the sexual preferences of Facebook users from pages they “like”.

“Even if you think you’re keeping your information private, we can learn a lot about you,” said Jennifer Golbeck, a University of Maryland computer scientist who conducts research similar to the one used to identify Facebook users.

Statistically, it’s not hard to do. There are only so many short, 31-year-old Jewish writers in San Francisco (actually, come to think of it, there are probably a lot of people like me in this city). While an algorithm may not be able to identify every single person, it can dramatically narrow the search to the point where it’s easy for a determined person to find the information they need.

People may disagree about whether or not government agencies should have private information, but let’s not pretend they can’t learn anything they want from what information they have.

View the Original article

Ways to Identify and Tackle Click Fraud

Identifying and tracking examples of click fraud is the first step to eradicating the problem. Click fraud is an enormous drain on the resources of advertisers operating on a national and international scale, estimated to occupy around 30% of all pay per click advertising spend. With so much at stake, its no wonder the search engines are investing so much time and effort into devising solutions.



One way in which Search Engines and other pay per click programme providers have attempted to curb the growing click fraud problem is through introducing IP address repetition algorithms. These formulae are designed to pick up on suspicious click patterns emanating from a singular IP address, which can help to uncover the existence of click farms and competitor-led sabotage, as well as identifying potential fraudsters at source.



However, there is an array of problems with this method of attempting to identify the fraudsters. Firstly, fraudsters logging on through a dialup modem, DSL line or cable modem can almost completely bypass this check, as with every new online session, a new IP address is generated. Furthermore, there is an extensive range of software available to alter IP addresses, which again can be used for 'cheating' the algorithm. Cookie and session tracking are other methods by which search engines can attempt to uncover potential fraudulent activity, but again there are ways around these for the fraudsters.



More comprehensive software is being developed which profiles and reports on the browsing habits of each click-through to enable companies to track and monitor suspicious behavior, although this could be seen by many as intrusive and ineffective as anything on a small scale is still likely to go unnoticed, based on the vast coverage of ads across the internet.



The problem of click fraud recently hit the headlines with a class action raised against Google, prompting Google to offer $90million as a potential settlement. Perhaps an acceptance of their responsibilities, Google's offer goes some way to suggest the extent of click fraud, and its vast costs to the internet economy.



There are a number of self-help remedies that can be implemented to keep an organization out of trouble. The first of these remedies is the reliance on search engine optimization and organic listings. If a site is well and fully optimized, it could eventually realize a ranking that another site is willing to pay $2.50 a click for. Similarly, with organically high rankings there are no clickthrough rates, therefore the costs associated with PPC are not applicable. Although the process is significantly more laborious and takes significantly longer to see results, the SEO process is much cheaper in the long run, and with an estimated 25-30% of all clicks being performed fraudulently, an organically high listing can save money which would otherwise be drained by click fraud for more beneficial reinvestment.



Year on year, as the pay per click advertising market continues to grow and expand, surely click fraud will follow suit. Unless an effective means of preventing click fraud is developed and successfully implemented, buyers will steadily lose confidence in the advertising medium and turn to more effective, less wasteful marketing methods, which would seriously hit the search engines and could potentially threaten the online economy as a whole.

NSA reportedly leveraging Google cookies and leaked mobile location data to identify hacking targets

You know those cookies web services use to track your history and serve up personalized ads? It appears that the government is using them too. The National Security Agency is apparently leveraging a Google-specific cookie to tap into the computers of suspicious users, according to presentation slides Edward Snowden leaked to The Washington Post. With an assist from Mountain View's "PREF" file, the NSA can track a target's web visits, then identify the computer and send a remote exploit in.

What's more, the documents also show that the outfit has used a program called "HAPPYFOOT" to map internet addresses to precise physical locations leaked by mobile apps when generating locally-germane ads. Perhaps the key takeaway here is this newest revelation's scope: The government could gain alarmingly precise information about individuals using data already spread throughout the internet, seeded under the not-quite-as-threatening guise of marketing and/or social media. Consider this a friendly reminder to clear your browser history, courtesy of Uncle Sam.

View the Original article