1
00:00:06,560 --> 00:00:11,560
- In 19.4, we're gonna be
talking about biometrics.

2
00:00:12,023 --> 00:00:13,770
Now biometrics are physical,

3
00:00:13,770 --> 00:00:16,510
or behavioral human
characteristics that can be used

4
00:00:16,510 --> 00:00:18,870
to digitally identify a person.

5
00:00:18,870 --> 00:00:21,190
We talked about this in the last lesson.

6
00:00:21,190 --> 00:00:24,185
Now biometric options are
physiological markers,

7
00:00:24,185 --> 00:00:26,434
which are what you are,

8
00:00:26,434 --> 00:00:30,090
or behavioral traits which is what you do.

9
00:00:30,090 --> 00:00:32,480
So, let's look at some of
those physiological markers,

10
00:00:32,480 --> 00:00:34,230
and behavioral traits.

11
00:00:34,230 --> 00:00:38,240
Fingerprint, retinal that's
your eye, iris also your eye,

12
00:00:38,240 --> 00:00:43,240
your vein in your hand or your
palm, facial, voice print,

13
00:00:43,460 --> 00:00:45,540
handwriting and and gait,

14
00:00:45,540 --> 00:00:47,620
which is really just a
fancy way of saying walking.

15
00:00:49,160 --> 00:00:53,010
The fingerprint is the unique
patterns of finger ridges

16
00:00:53,010 --> 00:00:55,330
and valleys.

17
00:00:55,330 --> 00:00:57,820
Retinal is the unique structure
of capillaries that supply

18
00:00:57,820 --> 00:00:59,010
the retina with blood.

19
00:00:59,980 --> 00:01:03,790
The iris is your unique
color and iris patterns.

20
00:01:03,790 --> 00:01:07,450
And vein is unique vein
patterns in your palm

21
00:01:07,450 --> 00:01:09,280
and in your fingers.

22
00:01:09,280 --> 00:01:11,660
I think this is just so amazing
with the billions of people

23
00:01:11,660 --> 00:01:14,670
in the world, right, that
we have unique fingerprints,

24
00:01:14,670 --> 00:01:18,110
unique retinal, iris, and vein patterns.

25
00:01:18,980 --> 00:01:20,300
Then we have facial,

26
00:01:20,300 --> 00:01:23,840
which is identification
based on facial features,

27
00:01:23,840 --> 00:01:26,610
such as an arrangement of your
eyes, your nose, your mouth.

28
00:01:26,610 --> 00:01:27,990
And there's definitely still some issues

29
00:01:27,990 --> 00:01:31,090
with facial recognition in terms
of how it recognizes people

30
00:01:31,090 --> 00:01:32,950
of color, women, and children.

31
00:01:34,660 --> 00:01:37,530
Voice print is the unique vocal attributes

32
00:01:37,530 --> 00:01:42,273
including anatomical, pitch,
and speaking patterns.

33
00:01:42,273 --> 00:01:45,219
Your handwriting is an analysis
of the shape, the speed,

34
00:01:45,219 --> 00:01:47,990
the stroke, the pen
pressure, and the timing.

35
00:01:47,990 --> 00:01:50,153
And gait is your walking characteristics,

36
00:01:50,153 --> 00:01:50,990
such as step length,

37
00:01:50,990 --> 00:01:55,990
step width, walking speed,
even joint rotation, and angle.

38
00:01:56,099 --> 00:01:58,410
So, fingerprint, retinal, iris, vein,

39
00:01:58,410 --> 00:02:01,679
and facial are all physiological,
they're what you are.

40
00:02:01,679 --> 00:02:03,026
Voice print, handwriting,

41
00:02:03,026 --> 00:02:05,300
and gait are all behavioral, right,

42
00:02:05,300 --> 00:02:10,280
behavioral being what you do.

43
00:02:10,280 --> 00:02:13,980
Now, biometric authentication
offers several advantages

44
00:02:13,980 --> 00:02:17,550
over traditional authentication
methods like passwords

45
00:02:17,550 --> 00:02:19,920
and pins.

46
00:02:19,920 --> 00:02:21,830
Well, first off, biometric
markers and traits are unique

47
00:02:21,830 --> 00:02:24,960
to each individual, and they're
really difficult to forge.

48
00:02:24,960 --> 00:02:28,100
So, it provides a higher
level of security,

49
00:02:28,100 --> 00:02:32,750
and it helps prevent identity
theft or unauthorized access.

50
00:02:32,750 --> 00:02:37,520
And biometric authentication
is really convenient for users,

51
00:02:37,520 --> 00:02:40,000
as they don't need to remember anything,

52
00:02:40,000 --> 00:02:42,640
and they don't need to carry
any additional credentials

53
00:02:42,640 --> 00:02:45,890
like a token, or a smart card.

54
00:02:47,440 --> 00:02:49,550
The biometric adoption's been interesting.

55
00:02:49,550 --> 00:02:52,690
You know, it's traditionally
been influenced by cost,

56
00:02:52,690 --> 00:02:55,920
enrollment time, and user acceptance.

57
00:02:57,099 --> 00:03:00,379
And for a long time, people
were very, very hesitant

58
00:03:00,379 --> 00:03:02,770
of enrolling in a biometric system.

59
00:03:02,770 --> 00:03:05,420
There was just a, I don't want
you to have my information,

60
00:03:05,420 --> 00:03:07,410
it just felt a little creepy,

61
00:03:07,410 --> 00:03:09,610
and people just didn't
seem to want to do it.

62
00:03:09,610 --> 00:03:13,910
And then when Apple introduced
biometric recognition

63
00:03:13,910 --> 00:03:17,650
on their devices, first your
finger and then now your face,

64
00:03:17,650 --> 00:03:19,980
all of a sudden, adoption
just took right off.

65
00:03:19,980 --> 00:03:22,570
People all of a sudden
became much more comfortable

66
00:03:22,570 --> 00:03:24,260
with it.

67
00:03:24,260 --> 00:03:26,310
The first thing that needs
to happen when someone

68
00:03:26,310 --> 00:03:29,010
is going to be using a biometric system

69
00:03:29,010 --> 00:03:30,450
is that we have to enroll them.

70
00:03:30,450 --> 00:03:31,960
Now during the enrollment process,

71
00:03:31,960 --> 00:03:33,809
the biometric registration system

72
00:03:33,809 --> 00:03:35,740
will take multiple measurements

73
00:03:35,740 --> 00:03:39,290
and goes through a series
of validation processes.

74
00:03:39,290 --> 00:03:40,930
Now those measurements are either hashed,

75
00:03:40,930 --> 00:03:45,930
or encrypted and they need to
be obviously stored securely.

76
00:03:46,370 --> 00:03:48,870
But of note is that biometric systems

77
00:03:48,870 --> 00:03:53,329
may inadvertently detect drug
usage, illness, and pregnancy.

78
00:03:54,889 --> 00:03:57,530
And that facial recognition is error prone

79
00:03:57,530 --> 00:03:59,900
within certain racial and age groups.

80
00:04:01,770 --> 00:04:05,530
So, in a biometric system,
we get to fine tune it.

81
00:04:05,530 --> 00:04:06,950
And we have some options here.

82
00:04:06,950 --> 00:04:09,990
We can fine tune it for
errors versus sensitivity.

83
00:04:11,359 --> 00:04:15,510
What we really want to do is
find the sweet spot in terms

84
00:04:15,510 --> 00:04:16,610
of our user community.

85
00:04:16,610 --> 00:04:18,520
Now, we may have other situations

86
00:04:18,520 --> 00:04:21,250
where we look to have less secure, well,

87
00:04:21,250 --> 00:04:23,690
probably never less
security, but more security.

88
00:04:23,690 --> 00:04:25,360
So, what does that really mean?

89
00:04:25,360 --> 00:04:27,770
Well, there are two
things we need to consider

90
00:04:27,770 --> 00:04:30,900
in our biometric accuracy
and when we're fine tuning.

91
00:04:30,900 --> 00:04:32,590
One is security, right?

92
00:04:32,590 --> 00:04:33,890
How much security do we need?

93
00:04:33,890 --> 00:04:35,640
Which is really how sensitive

94
00:04:35,640 --> 00:04:38,559
do we want that biometric system to be?

95
00:04:38,559 --> 00:04:40,800
The second is user frustration.

96
00:04:41,820 --> 00:04:44,900
If our users are having
trouble logging in,

97
00:04:44,900 --> 00:04:46,469
how quickly are they
going to get frustrated

98
00:04:46,469 --> 00:04:48,820
and how much is that
going to be a problem?

99
00:04:48,820 --> 00:04:50,800
So, security and frustration,

100
00:04:50,800 --> 00:04:53,320
those are two things we're
going to focus in on.

101
00:04:53,320 --> 00:04:56,160
And we have two types of errors.

102
00:04:57,160 --> 00:05:00,200
We have what's known as
false reject rates or FRRs,

103
00:05:00,200 --> 00:05:02,870
also called type one errors.

104
00:05:02,870 --> 00:05:04,469
And in a type one error,

105
00:05:04,469 --> 00:05:06,910
what's happening is a user is legitimate,

106
00:05:06,910 --> 00:05:09,030
but when they go to authenticate,

107
00:05:09,030 --> 00:05:11,370
their credentials aren't being accepted.

108
00:05:11,370 --> 00:05:14,450
Whether it's a finger or eye
or facial, they're saying, no,

109
00:05:14,450 --> 00:05:16,760
that's not you.

110
00:05:16,760 --> 00:05:19,030
And that's because we have
fine-tuned the system to be

111
00:05:19,030 --> 00:05:20,719
so sensitive, right,

112
00:05:20,719 --> 00:05:23,580
that any little change
might make a difference.

113
00:05:24,440 --> 00:05:27,970
The second is false accept rates or FAR,

114
00:05:27,970 --> 00:05:30,280
which are type two errors.

115
00:05:30,280 --> 00:05:32,980
And that's when we have
detuned the system,

116
00:05:32,980 --> 00:05:37,260
it is so desensitized
that we accept someone

117
00:05:37,260 --> 00:05:38,469
who's kind of close, right?

118
00:05:38,469 --> 00:05:40,760
It doesn't have to be an exact match.

119
00:05:40,760 --> 00:05:43,600
And so, we have these false exceptions.

120
00:05:43,600 --> 00:05:48,130
Now, where they match, we
can see here in the middle,

121
00:05:48,130 --> 00:05:51,660
where they match is referred
to as the crossover error rate.

122
00:05:51,660 --> 00:05:53,210
And that's generally the sweet spot

123
00:05:53,210 --> 00:05:56,219
that in most organizations
we're going to tune to,

124
00:05:56,219 --> 00:05:58,610
where our false reject
rate or our type one errors

125
00:05:58,610 --> 00:06:01,500
and our false accept rate
or type two errors cross.

126
00:06:01,500 --> 00:06:03,540
Again, that's known as
a crossover error rate.

127
00:06:04,500 --> 00:06:06,190
Now, that's not always going to be true.

128
00:06:06,190 --> 00:06:07,930
There may be some times, perhaps,

129
00:06:07,930 --> 00:06:10,010
in your data center that you're willing

130
00:06:10,010 --> 00:06:12,240
to have a little bit more user frustration

131
00:06:12,240 --> 00:06:14,020
to have a higher level of security.

132
00:06:15,710 --> 00:06:17,580
Now, we really need to make sure

133
00:06:17,580 --> 00:06:20,720
that we are securely
storing biometric data.

134
00:06:20,720 --> 00:06:23,020
Because biometric data, if compromised,

135
00:06:23,020 --> 00:06:24,620
can't be changed like a password.

136
00:06:24,620 --> 00:06:27,270
I can't give you a new finger,
I can't give you a new eye.

137
00:06:27,270 --> 00:06:28,790
I can't give you a new face.

138
00:06:28,790 --> 00:06:31,599
So biometric data must be securely stored.

139
00:06:32,639 --> 00:06:34,720
Now since biometric data is used

140
00:06:34,720 --> 00:06:37,330
for uniquely identifying an individual,

141
00:06:37,330 --> 00:06:39,050
it should be considered PII,

142
00:06:39,050 --> 00:06:41,380
personally identifiable information,

143
00:06:41,380 --> 00:06:44,470
and it does require privacy
protection in accordance

144
00:06:44,470 --> 00:06:46,929
with applicable laws and regulations.

145
00:06:48,240 --> 00:06:52,139
Additionally, any information
revealed during enrollment,

146
00:06:52,139 --> 00:06:54,119
or while in use such as pregnancy,

147
00:06:54,119 --> 00:06:56,359
or illness must be kept confidential.

148
00:06:57,930 --> 00:07:01,000
That my friends brings us
to a three-second challenge.

149
00:07:01,000 --> 00:07:03,020
Five challenge questions,
three seconds each.

150
00:07:03,020 --> 00:07:05,140
Let's do it.

151
00:07:05,140 --> 00:07:07,280
The biometric what you are.

152
00:07:07,280 --> 00:07:09,319
What is that?

153
00:07:09,319 --> 00:07:09,773
One, two, three.

154
00:07:11,049 --> 00:07:13,740
That's gonna be a physiological marker.

155
00:07:13,740 --> 00:07:16,573
Number two, biometric what you do?

156
00:07:17,530 --> 00:07:19,120
One, two, three.

157
00:07:19,980 --> 00:07:21,339
That's a behavioral trait.

158
00:07:22,750 --> 00:07:25,080
Number three, the intersection

159
00:07:25,080 --> 00:07:27,860
of FRR, that's false reject rate,

160
00:07:27,860 --> 00:07:30,240
and FAR, that's false accept rate,

161
00:07:30,240 --> 00:07:32,669
what is that intersection
or that sweet spot called?

162
00:07:33,680 --> 00:07:35,280
One, two, three.

163
00:07:36,410 --> 00:07:37,990
And that was your crossover error rate,

164
00:07:37,990 --> 00:07:39,000
or your CER.

165
00:07:40,860 --> 00:07:44,500
Number four, a behavioral
trait that includes pitch

166
00:07:44,500 --> 00:07:46,333
and speaking patterns.

167
00:07:47,340 --> 00:07:51,757
One, two, three, that's
going to be your voice print,

168
00:07:53,050 --> 00:07:55,300
or your voice recognition.

169
00:07:56,230 --> 00:08:00,170
And number five, last one,
physiological marker that relies

170
00:08:00,170 --> 00:08:02,799
on the unique structure of capillaries.

171
00:08:04,309 --> 00:08:08,620
One, two, three, and that's
going to be a retinal scan.

172
00:08:09,930 --> 00:08:12,800
All right, that brings us
to a security in action

173
00:08:12,800 --> 00:08:14,929
about biometric tuning.

174
00:08:14,929 --> 00:08:17,820
And I may have given you a
hint to this one already.

175
00:08:17,820 --> 00:08:21,940
Your organization is implementing
biometric authentication.

176
00:08:21,940 --> 00:08:25,090
End user systems will
have facial recognition.

177
00:08:25,090 --> 00:08:28,750
The data center entry
will have palm scanning.

178
00:08:28,750 --> 00:08:33,130
Now, the goal is to balance
user acceptance and security.

179
00:08:33,130 --> 00:08:36,500
So, what would you recommend
using for tuning options?

180
00:08:36,500 --> 00:08:37,650
We really have two different types

181
00:08:37,650 --> 00:08:39,880
of markers we're going
to be tuning for, right?

182
00:08:39,880 --> 00:08:41,889
One is for facial recognition

183
00:08:41,889 --> 00:08:45,570
and the other is for palm
scanning, end-users facial,

184
00:08:45,570 --> 00:08:47,580
data center, palm.

185
00:08:47,580 --> 00:08:49,200
So, the question to you is,

186
00:08:49,200 --> 00:08:51,770
what would you recommend
for tuning options?

187
00:08:53,460 --> 00:08:55,840
Go ahead and put me on pause.

188
00:08:55,840 --> 00:08:57,729
Come back with your
tuning recommendations.

189
00:09:01,080 --> 00:09:03,760
Well, the first step is to
define the security requirements

190
00:09:03,760 --> 00:09:05,320
and the tolerance for error.

191
00:09:06,530 --> 00:09:10,179
For end-user systems, in
order to balance security

192
00:09:10,179 --> 00:09:13,330
and user acceptance, you would
probably recommend tuning

193
00:09:13,330 --> 00:09:16,100
to the CER, the crossover error rate.

194
00:09:16,100 --> 00:09:19,030
That's where the FRR and
the FAR kind of cross.

195
00:09:19,030 --> 00:09:20,450
That's that middle sweet spot.

196
00:09:21,410 --> 00:09:23,470
Now, you're going to do the same thing

197
00:09:23,470 --> 00:09:24,220
for the data center though?

198
00:09:24,220 --> 00:09:25,642
Well for the data center,

199
00:09:25,642 --> 00:09:27,469
one can assume that there is a need

200
00:09:27,469 --> 00:09:30,711
to enhance security and a higher tolerance

201
00:09:30,711 --> 00:09:31,836
for type one errors,

202
00:09:31,836 --> 00:09:33,670
false reject rate errors,

203
00:09:33,670 --> 00:09:36,580
in which case you would
probably recommend tuning

204
00:09:36,580 --> 00:09:38,460
to a greater sensitivity level.

205
00:09:39,370 --> 00:09:41,810
So really looking at your
environment and saying, okay,

206
00:09:41,810 --> 00:09:43,500
this is what my end users need to get

207
00:09:43,500 --> 00:09:45,510
into their end user system.

208
00:09:45,510 --> 00:09:48,270
But boy, we probably need a
different level of security

209
00:09:48,270 --> 00:09:50,340
for getting into our data center.

210
00:09:50,340 --> 00:09:52,990
And so, tuning each of
those appropriately.

211
00:09:52,990 --> 00:09:55,790
Doing that, definitely security in action.

212
00:09:55,790 --> 00:09:58,180
There's your word cloud.

213
00:09:58,180 --> 00:09:58,180
This is a really big one.

214
00:09:58,180 --> 00:09:59,470
So, take your time, go through it,

215
00:09:59,470 --> 00:10:01,300
make sure you recognize all the terms

216
00:10:01,300 --> 00:10:03,510
and concepts before moving on.

217
00:10:03,510 --> 00:10:05,110
And when you're ready,
head to the next lesson.

218
00:10:05,110 --> 00:10:06,830
I'll be waiting for you there.