Entries Tagged "biometrics"

Page 4 of 15

Fooling Facial Recognition Systems

This is some interesting research. You can fool facial recognition systems by wearing glasses printed with elements of other people’s faces.

Mahmood Sharif, Sruti Bhagavatula, Lujo Bauer, and Michael K. Reiter, “Accessorize to a Crime: Real and Stealthy Attacks on State-of-the-Art Face Recognition“:

ABSTRACT: Machine learning is enabling a myriad innovations, including new algorithms for cancer diagnosis and self-driving cars. The broad use of machine learning makes it important to understand the extent to which machine-learning algorithms are subject to attack, particularly when used in applications where physical security or safety is at risk. In this paper, we focus on facial biometric systems, which are widely used in surveillance and access control. We define and investigate a novel class of attacks: attacks that are physically realizable and inconspicuous, and allow an attacker to evade recognition or impersonate another individual. We develop a systematic method to automatically generate such attacks, which are realized through printing a pair of eyeglass frames. When worn by the attacker whose image is supplied to a state-of-the-art face-recognition algorithm, the eyeglasses allow her to evade being recognized or to impersonate another individual. Our investigation focuses on white-box face-recognition systems, but we also demonstrate how similar techniques can be used in black-box scenarios, as well as to avoid face detection.

News articles.

Posted on November 11, 2016 at 7:31 AMView Comments

Using Neural Networks to Identify Blurred Faces

Neural networks are good at identifying faces, even if they’re blurry:

In a paper released earlier this month, researchers at UT Austin and Cornell University demonstrate that faces and objects obscured by blurring, pixelation, and a recently-proposed privacy system called P3 can be successfully identified by a neural network trained on image datasets­ — in some cases at a more consistent rate than humans.

“We argue that humans may no longer be the ‘gold standard’ for extracting information from visual data,” the researchers write. “Recent advances in machine learning based on artificial neural networks have led to dramatic improvements in the state of the art for automated image recognition. Trained machine learning models now outperform humans on tasks such as object recognition and determining the geographic location of an image.”

Research paper

Posted on September 27, 2016 at 9:39 AMView Comments

Using Wi-Fi Signals to Identify People by Body Shape

Another paper on using Wi-Fi for surveillance. This one is on identifying people by their body shape. “FreeSense:Indoor Human Identification with WiFi Signals“:

Abstract: Human identification plays an important role in human-computer interaction. There have been numerous methods proposed for human identification (e.g., face recognition, gait recognition, fingerprint identification, etc.). While these methods could be very useful under different conditions, they also suffer from certain shortcomings (e.g., user privacy, sensing coverage range). In this paper, we propose a novel approach for human identification, which leverages WIFI signals to enable non-intrusive human identification in domestic environments. It is based on the observation that each person has specific influence patterns to the surrounding WIFI signal while moving indoors, regarding their body shape characteristics and motion patterns. The influence can be captured by the Channel State Information (CSI) time series of WIFI. Specifically, a combination of Principal Component Analysis (PCA), Discrete Wavelet Transform (DWT) and Dynamic Time Warping (DTW) techniques is used for CSI waveform-based human identification. We implemented the system in a 6m*5m smart home environment and recruited 9 users for data collection and evaluation. Experimental results indicate that the identification accuracy is about 88.9% to 94.5% when the candidate user set changes from 6 to 2, showing that the proposed human identification method is effective in domestic environments.

EDITED TO ADD (9/13): Related paper.

Posted on August 30, 2016 at 12:57 PMView Comments

Apple Patents Collecting Biometric Information Based on Unauthorized Device Use

Apple applied for a patent earlier this year on collecting biometric information of an unauthorized device user. The obvious application is taking a copy of the fingerprint and photo of someone using a stolen smartphone.

Note that I have no opinion on whether this is a patentable idea or the patent is valid.

EDITED TO ADD (9/13): There is potential prior art in the comments.

Posted on August 29, 2016 at 6:27 AMView Comments

The Fallibility of DNA Evidence

This is a good summary article on the fallibility of DNA evidence. Most interesting to me are the parts on the proprietary algorithms used in DNA matching:

William Thompson points out that Perlin has declined to make public the algorithm that drives the program. “You do have a black-box situation happening here,” Thompson told me. “The data go in, and out comes the solution, and we’re not fully informed of what happened in between.”

Last year, at a murder trial in Pennsylvania where TrueAllele evidence had been introduced, defense attorneys demanded that Perlin turn over the source code for his software, noting that “without it, [the defendant] will be unable to determine if TrueAllele does what Dr. Perlin claims it does.” The judge denied the request.

[…]

When I interviewed Perlin at Cybergenetics headquarters, I raised the matter of transparency. He was visibly annoyed. He noted that he’d published detailed papers on the theory behind TrueAllele, and filed patent applications, too: “We have disclosed not the trade secrets of the source code or the engineering details, but the basic math.”

It’s the same problem as any biometric: we need to know the rates of both false positives and false negatives. And if these algorithms are being used to determine guilt, we have a right to examine them.

EDITED TO ADD (6/13): Three more articles.

Posted on May 31, 2016 at 1:04 PMView Comments

Identifying People from their Driving Patterns

People can be identified from their “driver fingerprint“:

…a group of researchers from the University of Washington and the University of California at San Diego found that they could “fingerprint” drivers based only on data they collected from internal computer network of the vehicle their test subjects were driving, what’s known as a car’s CAN bus. In fact, they found that the data collected from a car’s brake pedal alone could let them correctly distinguish the correct driver out of 15 individuals about nine times out of ten, after just 15 minutes of driving. With 90 minutes driving data or monitoring more car components, they could pick out the correct driver fully 100 percent of the time.

The paper: “Automobile Driver Fingerprinting,” by Miro Enev, Alex Takahuwa, Karl Koscher, and Tadayoshi Kohno.

Abstract: Today’s automobiles leverage powerful sensors and embedded computers to optimize efficiency, safety, and driver engagement. However the complexity of possible inferences using in-car sensor data is not well understood. While we do not know of attempts by automotive manufacturers or makers of after-market components (like insurance dongles) to violate privacy, a key question we ask is: could they (or their collection and later accidental leaks of data) violate a driver’s privacy? In the present study, we experimentally investigate the potential to identify individuals using sensor data snippets of their natural driving behavior. More specifically we record the in-vehicle sensor data on the controller area-network (CAN) of a typical modern vehicle (popular 2009 sedan) as each of 15 participants (a) performed a series of maneuvers in an isolated parking lot, and (b) drove the vehicle in traffic along a defined ~50 mile loop through the Seattle metropolitan area. We then split the data into training and testing sets, train an ensemble of classifiers, and evaluate identification accuracy of test data queries by looking at the highest voted candidate when considering all possible one-vs-one comparisons. Our results indicate that, at least among small sets, drivers are indeed distinguishable using only in car sensors. In particular, we find that it is possible to differentiate our 15 drivers with 100% accuracy when training with all of the available sensors using 90% of driving data from each person. Furthermore, it is possible to reach high identification rates using less than 8 minutes of training data. When more training data is available it is possible to reach very high identification using only a single sensor (e.g., the brake pedal). As an extension, we also demonstrate the feasibility of performing driver identification across multiple days of data collection.

Posted on May 30, 2016 at 10:10 AMView Comments

Stealing Fingerprints

The news from the Office of Personnel Management hack keeps getting worse. In addition to the personal records of over 20 million US government employees, we’ve now learned that the hackers stole fingerprint files for 5.6 million of them.

This is fundamentally different from the data thefts we regularly read about in the news, and should give us pause before we entrust our biometric data to large networked databases.

There are three basic kinds of data that can be stolen. The first, and most common, is authentication credentials. These are passwords and other information that allows someone else access into our accounts and — usually — our money. An example would be the 56 million credit card numbers hackers stole from Home Depot in 2014, or the 21.5 million Social Security numbers hackers stole in the OPM breach. The motivation is typically financial. The hackers want to steal money from our bank accounts, process fraudulent credit card charges in our name, or open new lines of credit or apply for tax refunds.

It’s a huge illegal business, but we know how to deal with it when it happens. We detect these hacks as quickly as possible, and update our account credentials as soon as we detect an attack. (We also need to stop treating Social Security numbers as if they were secret.)

The second kind of data stolen is personal information. Examples would be the medical data stolen and exposed when Sony was hacked in 2014, or the very personal data from the infidelity website Ashley Madison stolen and published this year. In these instances, there is no real way to recover after a breach. Once the data is public, or in the hands of an adversary, it’s impossible to make it private again.

This is the main consequence of the OPM data breach. Whoever stole the data — we suspect it was the Chinese — got copies the security-clearance paperwork of all those government employees. This documentation includes the answers to some very personal and embarrassing questions, and now opens these employees up to blackmail and other types of coercion.

Fingerprints are another type of data entirely. They’re used to identify people at crime scenes, but increasingly they’re used as an authentication credential. If you have an iPhone, for example, you probably use your fingerprint to unlock your phone. This type of authentication is increasingly common, replacing a password — something you know — with a biometric: something you are. The problem with biometrics is that they can’t be replaced. So while it’s easy to update your password or get a new credit card number, you can’t get a new finger.

And now, for the rest of their lives, 5.6 million US government employees need to remember that someone, somewhere, has their fingerprints. And we really don’t know the future value of this data. If, in twenty years, we routinely use our fingerprints at ATM machines, that fingerprint database will become very profitable to criminals. If fingerprints start being used on our computers to authorize our access to files and data, that database will become very profitable to spies.

Of course, it’s not that simple. Fingerprint readers employ various technologies to prevent being fooled by fake fingers: detecting temperature, pores, a heartbeat, and so on. But this is an arms race between attackers and defenders, and there are many ways to fool fingerprint readers. When Apple introduced its iPhone fingerprint reader, hackers figured out how to fool it within days, and have continued to fool each new generation of phone readers equally quickly.

Not every use of biometrics requires the biometric data to be stored in a central server somewhere. Apple’s system, for example, only stores the data locally: on your phone. That way there’s no central repository to be hacked. And many systems don’t store the biometric data at all, only a mathematical function of the data that can be used for authentication but can’t be used to reconstruct the actual biometric. Unfortunately, OPM stored copies of actual fingerprints.

Ashley Madison has taught us all the dangers of entrusting our intimate secrets to a company’s computers and networks, because once that data is out there’s no getting it back. All biometric data, whether it be fingerprints, retinal scans, voiceprints, or something else, has that same property. We should be skeptical of any attempts to store this data en masse, whether by governments or by corporations. We need our biometrics for authentication, and we can’t afford to lose them to hackers.

This essay previously appeared on Motherboard.

Posted on October 2, 2015 at 6:35 AMView Comments

Face Recognition by Thermal Imaging

New research can identify a person by reading their thermal signature in complete darkness and then matching it with ordinary photographs.

Research paper:

Abstract: Cross modal face matching between the thermal and visible spectrum is a much desired capability for night-time surveillance and security applications. Due to a very large modality gap, thermal-to-visible face recognition is one of the most challenging face matching problem. In this paper, we present an approach to bridge this modality gap by a significant margin. Our approach captures the highly non-linear relationship be- tween the two modalities by using a deep neural network. Our model attempts to learn a non-linear mapping from visible to thermal spectrum while preserving the identity in- formation. We show substantive performance improvement on a difficult thermal-visible face dataset. The presented approach improves the state-of-the-art by more than 10% in terms of Rank-1 identification and bridge the drop in performance due to the modality gap by more than 40%.

Posted on August 5, 2015 at 6:02 AMView Comments

Yet Another New Biometric: Brainprints

New research:

In “Brainprint,” a newly published study in academic journal Neurocomputing, researchers from Binghamton University observed the brain signals of 45 volunteers as they read a list of 75 acronyms, such as FBI and DVD. They recorded the brain’s reaction to each group of letters, focusing on the part of the brain associated with reading and recognizing words, and found that participants’ brains reacted differently to each acronym, enough that a computer system was able to identify each volunteer with 94 percent accuracy. The results suggest that brainwaves could be used by security systems to verify a person’s identity.

I have no idea what the false negatives are, or how robust this biometric is over time, but the article makes the important point that unlike most biometrics this one can be updated.

“If someone’s fingerprint is stolen, that person can’t just grow a new finger to replace the compromised fingerprint — the fingerprint for that person is compromised forever. Fingerprints are ‘non-cancellable.’ Brainprints, on the other hand, are potentially cancellable. So, in the unlikely event that attackers were actually able to steal a brainprint from an authorized user, the authorized user could then ‘reset’ their brainprint,” Laszlo said.

Presumably the resetting involves a new set of acronyms.

Author’s self-archived version of the paper (pdf).

Posted on June 4, 2015 at 10:36 AMView Comments

Sidebar photo of Bruce Schneier by Joe MacInnis.