Slashdot Mirror
Ears Might Be Better Than Fingerprints For ID
An anonymous reader writes "A new study says that outer ear could be better unique identification mark in human beings than finger prints. 'When you're born your ear is fully formed. The lobe descends a little, but overall it stays the same. It's a great way to identify people,' said Mark Nixon, a computer scientist at the University of Southampton and leader of the research. Nixon and his team presented a paper at the IEEE Fourth International Conference on Biometrics and using an algorithm identified people with 99.6 per cent accuracy."
An anonymous reader adds a link to Wired's story on the same conference presentation, which adds this skeptical note: "'I have seen no scientific proof that the ear doesn’t change significantly over time. People tend to believe notions like these, and they are repeated over time,' said Anil Jain, a computer scientist at Michigan State University who was not involved in the study. 'Fingerprinting has a history of 100 years showing that it works, unless you destroy your fingerprints or work in an industry that gives you calluses.'"
Yeah but how often do you leave earprints at the scene of a crime?
So I guess the Ferengi have made first contact with us poor terrans and have begun influencing our culture...
I mean no biometric ID is ever likely to be 100%. What you are just changes over time so even if we could measure it perfectly, there has to be fudge factor built in. Then there are situations like wins and so on.
However, that's ok, it doesn't need to be perfect. Biometrics shouldn't be security on its own, it should be in tandem with a passcode and/or a key or the like. The idea isn't that any of it is perfect, of course not, just that trying to successfully break more than one is really hard. Like if a door just has a passcode, well then what someone has to do is find out a legit passcode and use it. Not too hard in theory at least. However if that passcode is tied to a fingerprint, well then that is a problem. Even if it is only 99% accurate that means you have to find the 1 person in 100 that will work with that particular passcode. That is near impossible.
The big problem with biometrics at this point doesn't really seem to be accuracy but spoofing. Now that isn't as large a problem as it may seem since it isn't like getting a fingerprint from someone and making a replica is the easiest thing in the world, but it is a much bigger problem than accuracy. So unless this method is much harder to spoof, I don't really see how it matters that much.
Vincent Van Gogh would beg to differ with you.
Genitalia Biometrics. TSA would be hitting two birds with one stone. Once they make sure there are no bombs around your pecker (or peckette), they match your pecker against a database of peckerheads. Genitalia are not known to change over time (except my wife's), so they would probably need an If IsWife() { RaiseToleranceThreshold(); }; to prevent false positives (but not HIV positives, still need condoms for that).
'We are trying to prove ourselves wrong as quickly as possible, because only in that way can we find progress.' RPF
Have you ever seen people with jewelry that stretches their ears in a significant ways? What about wrestlers? Some of these peoples ears bare little resemblance to what they did when born. Now granted people can burn their finger tips and do all kinds of other crap as well, but this kind of mutilation is usually intentional as compared to the examples above (yes... I know people can lose fingers to a saw too...)
The martial arts crowd would be pretty immune to unique profiles, their ears develop pretty homogeneously with their career.
'We are trying to prove ourselves wrong as quickly as possible, because only in that way can we find progress.' RPF
> unless you destroy your fingerprints
Having inadvertantly taken my fingerprints off one hand at one point (yes, it was VERY painful, thank you), and found (as many others have) that they grow back... can you actually damage them so bad/repeatedly they don't grow back, and still have things like, erm, fingertips?
No more earmarks!
"Fingerprinting has a history of 100 years showing that it works."
Fingerprinting has a history of well over 100 years, but what we see is that it works as long as it is not seriously challenged. In its only major rigorous challenge, the 50Kx50k text, substantial problems emerged.
Keep in mind that fingerprints are never admitted into evidence, never used for identification, never even examined. Never. A finger touches a surface and it leaves a partial copy. An investigator finds it and puts powder (matrix) on it, which creates a visible picture of the copy. It is often not possible to get a good photo of the copy, so someone uses tape or other gear to get an image of the picture of the copy. Then someone photographs the tape containing the image of the picture of the copy. Then a print of the photograph of the tape of the image of the picture of the copy is created. If there are no more steps, which would be unusual, that print is what is actually used for evidence or analysis. Scientifically-minded readers will have already tallied up at least a partial list of the errors introduced at each step of the process.
And what sort of analysis is done? The best lab in the country, the FBI, uses an analysis process taught by a high school grad who washed out of college after two years. Obviously, other labs do not enjoy such high standards. What standards do they use, you may ask? None. There are no required national standards for fingerprint analysts. There are guidelines that suggest that a high school diploma should be required, but the advisory guidelines bind no one.
But at least they use a rigorous process with well-defined standards?
"The International Association for Identification assembled in its 58th annual conference... based on a three-year study by its Standardization Committee, hereby states that no valid basis exists at this time for requiring that a predetermined minimum of friction ridge [fingerprint] characteristcs must be present in two impressions in order to establish positive identification."
So no, there are no standards, which is a good thing because the relevant international body has determined that there is "no valid basis" for establishing one.
So now they say that they can get better results by looking at someone's ears? Hm... Well, the good news is that they're probably right. The bad news is that they've got a long way to go before they can say that it's any great accomplishment.
and Evander Holyfield.
Game: Player 'Donald J Trump' now has AI skill level 'experimental'.
I was born with ears that stuck out worse that Prince Charles. I was teased about them all through school.
In college I had my ears "tucked," which basically made them lay flat against my head. I had generous grandparents.
Anyway, the point is that to do this, (the following not for the queasy), they slice open your ear, take out the cartilage (which is what forms all the unique bumps and curves of your ear), manually reshape it, stick it back in, and then sew you up.
Not only did my ears finally not stick out, but they looked totally different than they did before: none of the curves matched, and even my earlobes are a different shape (the bottoms are trimmed a bit and then stitched back to your head.)
This is not terribly expensive surgery, and while a bit painful, if I were a criminal trying to beat a set of "earprints" somehow left at the scene of a crime, I'd have it done in a second.
What you described is the classic difference between sensitivity and positive predictive value (PPV). Sensitivity is a basic characteristic of a test, in this case 99.6% (Actually the TFA mentioned accuracy, which is a bit different, but let's not nitpick). PPV tells us what is the chance that a positive result (in this case, an ear match), is a true positive. Since the equation is TP/(TP+FP) (TP True positive; FP - False positive), it is affected by how common (or rare) the trait we are looking for is in the population we are checking. Since a terrorist is a rare occurrance, the PPV is (very) low.
However, if we change the test a bit we can improve the PPV. Let's say we do not use the ears as a single test, but rather as a verification for the ID. A person shows a passport and then his ears are compared to what is stored at the computer. Here the test is used just for verification and not identification and we have a much better PPV (In this case a positive is actually a mismatch between the passport and ears) and the system can be used to detect people with fake IDs.
BTW, this is used in many places where fingerprints are used. I don't know about other countries, but in Israel citizens can register their fingerprints and bypass passport control by going to a booth where you pass a magnetic card (containing your ID) and then you put your fingerprints for verification.
So is 99.6% good enough? depending on the application. Oh, does anyone know what is the accuracy of fingerprint recognition devices?
Whenever in an argument, remember this.