WEBVTT

1
00:00:00.360 --> 00:00:01.740
In this lesson,

2
00:00:01.740 --> 00:00:05.610
we will learn about Collaborative Considerations.

3
00:00:05.610 --> 00:00:08.970
Collaborative considerations involve the methods

4
00:00:08.970 --> 00:00:13.800
and protocols that enable secure, cooperative computing

5
00:00:13.800 --> 00:00:17.340
and data sharing among multiple parties

6
00:00:17.340 --> 00:00:20.280
without compromising security.

7
00:00:20.280 --> 00:00:24.030
Collaborative considerations include techniques

8
00:00:24.030 --> 00:00:26.730
such as homomorphic encryption

9
00:00:26.730 --> 00:00:30.180
and secure multi-party computation.

10
00:00:30.180 --> 00:00:33.330
Homomorphic encryption is a technique

11
00:00:33.330 --> 00:00:35.280
that enables computations

12
00:00:35.280 --> 00:00:38.190
to be performed on encrypted data

13
00:00:38.190 --> 00:00:41.580
without ever needing to decrypt the data.

14
00:00:41.580 --> 00:00:46.170
Secure multi-party computation or SMPC

15
00:00:46.170 --> 00:00:48.210
is a cryptographic method

16
00:00:48.210 --> 00:00:51.330
that allows several parties to work together

17
00:00:51.330 --> 00:00:56.330
to compute a result using their own individual inputs,

18
00:00:56.550 --> 00:00:59.970
while keeping each party's individual input

19
00:00:59.970 --> 00:01:01.770
completely private.

20
00:01:01.770 --> 00:01:05.130
Let's learn more about homomorphic encryption

21
00:01:05.130 --> 00:01:08.430
and secure multi-party computation.

22
00:01:08.430 --> 00:01:12.240
First, we have homomorphic encryption.

23
00:01:12.240 --> 00:01:16.350
Homomorphic encryption is a special type of encryption

24
00:01:16.350 --> 00:01:18.180
that allows computations

25
00:01:18.180 --> 00:01:22.890
to be performed on data without decrypting it first.

26
00:01:22.890 --> 00:01:27.660
This means an organization can process data securely

27
00:01:27.660 --> 00:01:32.430
even when outsourcing to a third party cloud provider.

28
00:01:32.430 --> 00:01:37.430
In this situation, the cloud provider can perform operations

29
00:01:37.650 --> 00:01:42.150
on the encrypted data, but cannot read or access

30
00:01:42.150 --> 00:01:47.040
the actual information because it remains encrypted.

31
00:01:47.040 --> 00:01:51.330
Homomorphic encryption differs from traditional encryption

32
00:01:51.330 --> 00:01:55.470
because it allows computation on the encrypted data

33
00:01:55.470 --> 00:01:58.920
without needing the secret decryption key.

34
00:01:58.920 --> 00:02:01.140
This keeps the data private

35
00:02:01.140 --> 00:02:04.260
and confidential during processing,

36
00:02:04.260 --> 00:02:06.540
making it especially valuable

37
00:02:06.540 --> 00:02:10.230
for industries like finance and healthcare

38
00:02:10.230 --> 00:02:14.190
where protecting sensitive information is crucial.

39
00:02:14.190 --> 00:02:18.960
Homomorphic encryption works by using a public key

40
00:02:18.960 --> 00:02:23.010
to encrypt data, and an algebraic system

41
00:02:23.010 --> 00:02:27.900
that enables mathematical operations on the encrypted data.

42
00:02:27.900 --> 00:02:32.460
The related private key is not provided to the organization

43
00:02:32.460 --> 00:02:35.370
that will analyze or process the data,

44
00:02:35.370 --> 00:02:38.400
meaning they can conduct data manipulation

45
00:02:38.400 --> 00:02:42.600
and analysis on the data, but they can't decrypt it.

46
00:02:42.600 --> 00:02:46.920
There are three main types of homomorphic encryption,

47
00:02:46.920 --> 00:02:49.230
partially homomorphic encryption,

48
00:02:49.230 --> 00:02:51.630
somewhat homomorphic encryption,

49
00:02:51.630 --> 00:02:54.810
and fully homomorphic encryption.

50
00:02:54.810 --> 00:02:59.430
Partially homomorphic encryption allows specific operations

51
00:02:59.430 --> 00:03:03.210
like addition or multiplication on the data.

52
00:03:03.210 --> 00:03:05.490
Somewhat homomorphic encryption

53
00:03:05.490 --> 00:03:08.220
supports a limited set of operations

54
00:03:08.220 --> 00:03:10.920
a certain amount of times.

55
00:03:10.920 --> 00:03:14.130
Finally, fully homomorphic encryption

56
00:03:14.130 --> 00:03:18.750
allows any type of computation on the encrypted data

57
00:03:18.750 --> 00:03:20.910
while keeping it secure.

58
00:03:20.910 --> 00:03:24.240
Homomorphic encryption relies on techniques

59
00:03:24.240 --> 00:03:26.910
like private information retrieval,

60
00:03:26.910 --> 00:03:29.340
secure function evaluation,

61
00:03:29.340 --> 00:03:32.250
and a private function evaluation

62
00:03:32.250 --> 00:03:36.330
to maintain data privacy during processing.

63
00:03:36.330 --> 00:03:40.020
Private information retrieval allows a user

64
00:03:40.020 --> 00:03:42.870
to retrieve data from a database

65
00:03:42.870 --> 00:03:46.230
without revealing what data is accessed.

66
00:03:46.230 --> 00:03:50.010
Secure function evaluation lets two parties

67
00:03:50.010 --> 00:03:51.990
compute a function together

68
00:03:51.990 --> 00:03:55.680
without sharing their individual inputs.

69
00:03:55.680 --> 00:03:58.620
Finally, private function evaluation

70
00:03:58.620 --> 00:04:02.370
goes a little bit further by allowing two parties

71
00:04:02.370 --> 00:04:04.740
to compute a private function

72
00:04:04.740 --> 00:04:09.740
without revealing either the inputs or the function itself.

73
00:04:10.050 --> 00:04:13.440
The main challenge of homomorphic encryption

74
00:04:13.440 --> 00:04:17.580
is that it is currently very slow, making it difficult

75
00:04:17.580 --> 00:04:20.400
to use in modern applications.

76
00:04:20.400 --> 00:04:23.880
Major companies like Microsoft and IBM

77
00:04:23.880 --> 00:04:27.690
are working to improve its speed and efficiency,

78
00:04:27.690 --> 00:04:32.130
aiming to make homomorphic encryption a practical tool

79
00:04:32.130 --> 00:04:35.760
for secure data processing in the future.

80
00:04:35.760 --> 00:04:40.200
Wow, homomorphic encryption is a complex topic.

81
00:04:40.200 --> 00:04:44.580
But for the exam, remember that homomorphic encryption

82
00:04:44.580 --> 00:04:47.400
allows encrypted data to be processed

83
00:04:47.400 --> 00:04:49.500
without being decrypted.

84
00:04:49.500 --> 00:04:53.490
The results of those computations can then be decrypted

85
00:04:53.490 --> 00:04:58.490
only by the original data using their private key,

86
00:04:58.590 --> 00:05:03.120
maintaining data security throughout the entire process.

87
00:05:03.120 --> 00:05:08.010
Second, we have secure multi-party computation.

88
00:05:08.010 --> 00:05:12.540
Secure multi-party computation or SMPC

89
00:05:12.540 --> 00:05:16.440
and homomorphic encryption, both aim to enable

90
00:05:16.440 --> 00:05:21.440
secure data processing without exposing private information,

91
00:05:21.780 --> 00:05:24.810
but they approach the problem differently.

92
00:05:24.810 --> 00:05:28.440
While homomorphic encryption allows computations

93
00:05:28.440 --> 00:05:30.120
on encrypted data,

94
00:05:30.120 --> 00:05:35.120
SMPC lets multiple parties jointly compute a result

95
00:05:35.280 --> 00:05:37.290
using their private inputs

96
00:05:37.290 --> 00:05:41.130
without revealing these inputs to each other.

97
00:05:41.130 --> 00:05:45.990
Secure multi-party computation is a cryptographic technique

98
00:05:45.990 --> 00:05:50.220
that enables multiple parties to jointly compute a function

99
00:05:50.220 --> 00:05:51.990
with their private data

100
00:05:51.990 --> 00:05:55.230
without revealing that data to each other.

101
00:05:55.230 --> 00:05:59.280
This approach ensures that participants can collaborate

102
00:05:59.280 --> 00:06:03.750
and reach a result without compromising the privacy

103
00:06:03.750 --> 00:06:06.510
of their individual inputs.

104
00:06:06.510 --> 00:06:09.120
Unlike traditional cryptography,

105
00:06:09.120 --> 00:06:13.560
which focuses on protecting data from external threats,

106
00:06:13.560 --> 00:06:16.470
secure multi-party computation

107
00:06:16.470 --> 00:06:20.700
protects sensitive information from the very participants

108
00:06:20.700 --> 00:06:23.520
involved in the computation.

109
00:06:23.520 --> 00:06:26.250
So imagine a group of companies

110
00:06:26.250 --> 00:06:30.510
that want to calculate the average of their sales data

111
00:06:30.510 --> 00:06:34.860
without revealing their individual numbers to each other.

112
00:06:34.860 --> 00:06:37.770
With secure multi-party computation,

113
00:06:37.770 --> 00:06:42.150
each company can input its sales into the computation,

114
00:06:42.150 --> 00:06:44.970
and the final result or the average

115
00:06:44.970 --> 00:06:47.040
can be shared with everyone

116
00:06:47.040 --> 00:06:51.300
without exposing any of the companies' private inputs.

117
00:06:51.300 --> 00:06:54.870
This is achieved through cryptographic protocols

118
00:06:54.870 --> 00:06:57.240
that ensure the data remains hidden

119
00:06:57.240 --> 00:07:00.570
during the entire computation process,

120
00:07:00.570 --> 00:07:05.520
allowing only the final outcome to be visible to everyone.

121
00:07:05.520 --> 00:07:09.000
Within secure multi-party computation,

122
00:07:09.000 --> 00:07:13.620
techniques like secure function evaluation or SFE

123
00:07:13.620 --> 00:07:17.550
and private function evaluation or PFE

124
00:07:17.550 --> 00:07:20.520
are used to enhance privacy.

125
00:07:20.520 --> 00:07:24.540
Secure function evaluation allows multiple parties

126
00:07:24.540 --> 00:07:27.660
to jointly evaluate a known function

127
00:07:27.660 --> 00:07:30.210
without revealing their inputs,

128
00:07:30.210 --> 00:07:35.010
ensuring that everyone only sees the final result.

129
00:07:35.010 --> 00:07:39.330
Private function evaluation takes this a step further,

130
00:07:39.330 --> 00:07:43.080
allowing parties to compute a private function

131
00:07:43.080 --> 00:07:47.970
without revealing the function itself or the inputs used,

132
00:07:47.970 --> 00:07:51.120
adding an extra layer of security.

133
00:07:51.120 --> 00:07:53.580
Overall, the primary strength

134
00:07:53.580 --> 00:07:57.510
of secure multi-party computation is its ability

135
00:07:57.510 --> 00:08:02.250
to maintain data privacy even in collaborative settings

136
00:08:02.250 --> 00:08:05.730
where multiple parties need to work together.

137
00:08:05.730 --> 00:08:10.730
This makes SMPC particularly useful in scenarios

138
00:08:10.920 --> 00:08:15.450
like joint data analysis, secure voting systems,

139
00:08:15.450 --> 00:08:19.050
or any situation where sensitive information

140
00:08:19.050 --> 00:08:21.060
must remain confidential

141
00:08:21.060 --> 00:08:24.540
while still contributing to a shared result.

142
00:08:24.540 --> 00:08:29.540
So remember, collaborative security techniques are essential

143
00:08:29.940 --> 00:08:33.930
for enabling secure computing and data sharing

144
00:08:33.930 --> 00:08:38.490
among multiple parties without compromising privacy.

145
00:08:38.490 --> 00:08:42.600
These techniques include homomorphic encryption

146
00:08:42.600 --> 00:08:47.310
and secure multi-party computation or SMPC,

147
00:08:47.310 --> 00:08:51.360
both of which allow the data to be processed securely

148
00:08:51.360 --> 00:08:54.750
without revealing sensitive information.

149
00:08:54.750 --> 00:08:59.070
Homomorphic encryption does this by allowing computations

150
00:08:59.070 --> 00:09:03.960
to be performed on the encrypted data without decrypting it,

151
00:09:03.960 --> 00:09:07.290
keeping the data private even during processing

152
00:09:07.290 --> 00:09:09.600
by third party providers.

153
00:09:09.600 --> 00:09:13.380
Secure multi-party computation, on the other hand,

154
00:09:13.380 --> 00:09:17.430
allows multiple parties to compute a result

155
00:09:17.430 --> 00:09:19.800
using their private inputs

156
00:09:19.800 --> 00:09:23.610
without exposing those inputs to one another,

157
00:09:23.610 --> 00:09:26.310
ensuring that privacy is maintained

158
00:09:26.310 --> 00:09:28.590
throughout the computation.

159
00:09:28.590 --> 00:09:32.700
These methods are used to maintain confidentiality

160
00:09:32.700 --> 00:09:37.170
and integrity, particularly in collaborative environments

161
00:09:37.170 --> 00:09:40.500
where sensitive information needs to be shared

162
00:09:40.500 --> 00:09:42.573
and analyzed securely.