1 00:00:00,680 --> 00:00:02,570 Biometric readers. 2 00:00:02,570 --> 00:00:04,750 Biometrics rely on physical characteristics 3 00:00:04,750 --> 00:00:07,180 to identify a person properly. 4 00:00:07,180 --> 00:00:09,477 This is most commonly done by using your fingerprints, 5 00:00:09,477 --> 00:00:11,516 by scanning the retina inside your eye, 6 00:00:11,516 --> 00:00:13,090 or by measuring the distance 7 00:00:13,090 --> 00:00:15,270 between different parts of your face. 8 00:00:15,270 --> 00:00:16,700 Now, if you remember all the way back 9 00:00:16,700 --> 00:00:17,970 to section one of this course, 10 00:00:17,970 --> 00:00:20,520 we talked about the five factors of authentication. 11 00:00:20,520 --> 00:00:22,748 This was something you know, something you have, 12 00:00:22,748 --> 00:00:25,000 something you are, something you do, 13 00:00:25,000 --> 00:00:26,430 and somewhere you are. 14 00:00:26,430 --> 00:00:27,900 Now, when we talk about biometrics, 15 00:00:27,900 --> 00:00:30,519 we're focused on that third factor: something you are. 16 00:00:30,519 --> 00:00:32,690 Because this is some part of you. 17 00:00:32,690 --> 00:00:35,740 It's your eye, it's your fingerprint, it's your voice, 18 00:00:35,740 --> 00:00:36,990 it's something that is innately 19 00:00:36,990 --> 00:00:39,487 part of your ability and part of your person. 20 00:00:39,487 --> 00:00:41,260 Now, when we talk about fingerprints, 21 00:00:41,260 --> 00:00:42,110 fingerprints have become 22 00:00:42,110 --> 00:00:44,240 a very common identification system. 23 00:00:44,240 --> 00:00:46,520 At this point, it's even gone beyond door locks, 24 00:00:46,520 --> 00:00:48,450 and it's now integrated into our smartphones 25 00:00:48,450 --> 00:00:49,756 and our laptops for log in. 26 00:00:49,756 --> 00:00:53,490 For example, if you have an iPhone between a 5s model 27 00:00:53,490 --> 00:00:55,160 all the way up through the eight series, 28 00:00:55,160 --> 00:00:58,039 those have a thumbprint login called touch ID. 29 00:00:58,039 --> 00:00:59,499 Whenever you press your index finger 30 00:00:59,499 --> 00:01:02,700 or your thumb to the sensor, it will log you in. 31 00:01:02,700 --> 00:01:05,700 Now, the newest iPhones have done away with touch ID 32 00:01:05,700 --> 00:01:07,100 in favor of face ID. 33 00:01:07,100 --> 00:01:09,010 So, if you have an iPhone X or newer, 34 00:01:09,010 --> 00:01:11,750 they actually have the front-facing camera scan your face 35 00:01:11,750 --> 00:01:13,220 and measure the distance between 36 00:01:13,220 --> 00:01:16,011 different areas of your face to uniquely identify you. 37 00:01:16,011 --> 00:01:18,560 This allows you to just hold up the phone in front 38 00:01:18,560 --> 00:01:20,000 and it logs you in automatically 39 00:01:20,000 --> 00:01:23,050 by identifying that it is you that's holding the phone. 40 00:01:23,050 --> 00:01:25,130 These types of devices are also being integrated, 41 00:01:25,130 --> 00:01:27,507 into door locks and physical access control systems, 42 00:01:27,507 --> 00:01:28,753 like man traps. 43 00:01:28,753 --> 00:01:31,160 I used to work in one high security area 44 00:01:31,160 --> 00:01:33,460 where every day you had to use a retina scan 45 00:01:33,460 --> 00:01:35,950 to gain access to the facility. 46 00:01:35,950 --> 00:01:37,730 I've also worked in places that have used 47 00:01:37,730 --> 00:01:39,380 a fingerprint and a PIN number 48 00:01:39,380 --> 00:01:41,060 that allowed you to get through a mantrap. 49 00:01:41,060 --> 00:01:43,570 It really does depend on how your system wants to be set up, 50 00:01:43,570 --> 00:01:46,660 and how your organization views its security practices. 51 00:01:46,660 --> 00:01:48,400 Now, with biometrics, we do have 52 00:01:48,400 --> 00:01:50,498 a couple of challenges that we have to focus on, 53 00:01:50,498 --> 00:01:53,740 namely, the acceptance and rejection rates. 54 00:01:53,740 --> 00:01:55,470 Obviously, since we're using biometrics, 55 00:01:55,470 --> 00:01:56,990 as a form of authentication, 56 00:01:56,990 --> 00:01:59,235 we should be worried about its false acceptance rate. 57 00:01:59,235 --> 00:02:01,257 The false acceptance rate or FAR 58 00:02:01,257 --> 00:02:04,980 is the rate that the system authenticates a user as valid, 59 00:02:04,980 --> 00:02:06,700 even though that person should not have been 60 00:02:06,700 --> 00:02:08,780 granted access to the system. 61 00:02:08,780 --> 00:02:11,360 For example, if you walked up to the fingerprint reader, 62 00:02:11,360 --> 00:02:13,560 placed your finger on it, and it accepts you, 63 00:02:13,560 --> 00:02:15,700 because the system thought you were me, 64 00:02:15,700 --> 00:02:18,240 that would be considered a false acceptance. 65 00:02:18,240 --> 00:02:19,530 To prevent unauthorized people 66 00:02:19,530 --> 00:02:21,272 from entering the building or using our system, 67 00:02:21,272 --> 00:02:24,310 we would want to ideally get that false acceptance rate, 68 00:02:24,310 --> 00:02:28,060 down to zero by increasing the sensitivity of our scanners. 69 00:02:28,060 --> 00:02:29,800 Now, on the other side of the spectrum, 70 00:02:29,800 --> 00:02:31,235 we have a false rejection. 71 00:02:31,235 --> 00:02:32,870 And a lot of people don't think, 72 00:02:32,870 --> 00:02:34,290 a false rejection is a problem 73 00:02:34,290 --> 00:02:37,233 but it is just as big of a problem as a false acceptance. 74 00:02:37,233 --> 00:02:39,015 Let's go back to the last example. 75 00:02:39,015 --> 00:02:41,726 If I increase the sensitivity of the fingerprint scanner 76 00:02:41,726 --> 00:02:42,995 to attempt to eliminate all 77 00:02:42,995 --> 00:02:45,670 of the false acceptance rates or FAR, 78 00:02:45,670 --> 00:02:49,754 I may inadvertently increase my false rejection rate or FRR. 79 00:02:49,754 --> 00:02:52,210 A false rejection occurs any time 80 00:02:52,210 --> 00:02:54,570 the biometrics system denies a user 81 00:02:54,570 --> 00:02:56,820 who should have been allowed access to the system. 82 00:02:56,820 --> 00:02:59,497 So, using that login example with my finger, 83 00:02:59,497 --> 00:03:01,160 let's assume that you were able to 84 00:03:01,160 --> 00:03:03,080 log in as me using your fingerprint. 85 00:03:03,080 --> 00:03:05,514 So, I increase that sensitivity up to its highest level. 86 00:03:05,514 --> 00:03:08,740 Now, there's no more false acceptances occurring. 87 00:03:08,740 --> 00:03:10,300 But about half of the time, 88 00:03:10,300 --> 00:03:12,570 when I try to log in, I'm being rejected 89 00:03:12,570 --> 00:03:14,610 even though I'm an authorized user. 90 00:03:14,610 --> 00:03:16,980 Now, this is the idea of a false rejection, 91 00:03:16,980 --> 00:03:19,590 and why it's also considered a big problem. 92 00:03:19,590 --> 00:03:21,150 Because if it's failing to allow me 93 00:03:21,150 --> 00:03:22,950 to authenticate half of the time, 94 00:03:22,950 --> 00:03:25,550 that means half the time, I can't get to work. 95 00:03:25,550 --> 00:03:27,700 So, how can I perfectly tune the system 96 00:03:27,700 --> 00:03:30,880 to eliminate the false acceptance and the false rejection? 97 00:03:30,880 --> 00:03:32,470 Well, I try to achieve a balance 98 00:03:32,470 --> 00:03:33,790 where the false acceptance rate 99 00:03:33,790 --> 00:03:36,530 and the false rejection rate become equal. 100 00:03:36,530 --> 00:03:39,750 This is known as the equal error rate or ERR, 101 00:03:39,750 --> 00:03:41,530 more commonly, though, you'll see this called 102 00:03:41,530 --> 00:03:44,750 the crossover error rate or CER out in the industry. 103 00:03:44,750 --> 00:03:47,350 This is because your false acceptance rate 104 00:03:47,350 --> 00:03:48,910 and your false rejection rate 105 00:03:48,910 --> 00:03:52,028 intersect at that crossover error rate or CER. 106 00:03:52,028 --> 00:03:54,503 The crossover error rate uses a measure 107 00:03:54,503 --> 00:03:57,192 of the effectiveness of a given biometrics system. 108 00:03:57,192 --> 00:03:59,130 So, when you're looking to purchase one, 109 00:03:59,130 --> 00:04:00,510 you can use this as a factor 110 00:04:00,510 --> 00:04:01,959 in your decision-making process. 111 00:04:01,959 --> 00:04:03,370 You want one that doesn't have 112 00:04:03,370 --> 00:04:05,229 a huge error to the positive side 113 00:04:05,229 --> 00:04:07,390 or a huge error to the negative side. 114 00:04:07,390 --> 00:04:10,220 If you can get one that has a good crossover error rate, 115 00:04:10,220 --> 00:04:11,700 that's going to make sure that your people 116 00:04:11,700 --> 00:04:13,410 are getting authenticated when you should be 117 00:04:13,410 --> 00:04:16,160 and rejected when they should be.