The document provides an overview of privacy-preserving data mining techniques in industry, focusing on challenges, lessons learned, and practical applications such as differential privacy. It discusses real-world privacy breaches and highlights Google's RAPPOR and Apple's on-device differential privacy as key case studies in maintaining user privacy. Key takeaways include the importance of understanding privacy techniques, the evolution of privacy breaches, and the development of methods that balance data utility with user privacy.

1. Privacy-preserving Data Mining
in Industry: Practical Challenges
and Lessons Learned
KDD 2018 Tutorial
August 2018
Krishnaram Kenthapadi (AI @ LinkedIn)
Ilya Mironov (Google AI)
Abhradeep Thakurta (UC Santa Cruz)
https://sites.google.com/view/kdd2018privacytutorial

2. Outline / Learning Outcomes
• Privacy breaches and lessons learned
• Evolution of privacy techniques
• Differential privacy: definition and techniques
• Privacy techniques in practice: Challenges and Lessons Learned
• Google’s RAPPOR
• Apple’s differential privacy deployment for iOS
• LinkedIn Salary: Privacy Design
• Key Takeaways

3. Privacy: A Historical Perspective
Evolution of Privacy Techniques and Privacy Breaches

4. Privacy Breaches and Lessons Learned
Attacks on privacy
•Governor of Massachusetts
•AOL
•Netflix
•Web browsing data
•Facebook
•Amazon
•Genomic data

5. born July 31, 1945
resident of 02138
Massachusetts Group Insurance Commission (1997):
Anonymized medical history of state employees (all
hospital visits, diagnosis, prescriptions)
Latanya Sweeney (MIT grad student): $20 – Cambridge
voter roll
William Weld vs Latanya Sweeney

6. 64
%uniquely identifiable with
ZIP + birth date + gender
(in the US population)
Golle, “Revisiting the Uniqueness of Simple Demographics in the US Population”,

7. Attacker's Advantage
Auxiliary information

8. August 4, 2006: AOL
Research publishes
anonymized search logs
of 650,000 users
August 9:
New York Times
AOL Data Release

9. Attacker's Advantage
Auxiliary information
Enough to succeed on a small fraction of inputs

10. Netflix Prize

11. Oct 2006: Netflix announces Netflix
Prize
• 10% of their users
• average 200 ratings per user
Narayanan, Shmatikov (2006):
Netflix Prize

12. Deanonymizing Netflix Data
Narayanan, Shmatikov, Robust De-
anonymization of Large Datasets (How to
Break Anonymity of the Netflix Prize
Dataset), 2008

13. ● Noam Chomsky in Our Times
● Farenheit 9/11
● Jesus of Nazareth
● Queer as Folk

14. Key idea:
● Similar intuition as the attack on medical records
● Medical records: Each person can be identified
based on a combination of a few attributes
● Web browsing history: Browsing history is unique for
each person
● Each person has a distinctive social network  links
appearing in one’s feed is unique
● Users likely to visit links in their feed with higher
probability than a random user
● “Browsing histories contain tell-tale marks of identity”
Su et al, De-anonymizing Web Browsing Data with Social Networks, 2017
De-anonymizing Web Browsing Data with Social Networks

15. Attacker's Advantage
Auxiliary information
Enough to succeed on a small fraction of inputs
High dimensionality

16. Ad targeting:
Korolova, “Privacy Violations Using Microtargeted Ads: A Case Study”, PADM
Privacy Attacks On Ad Targeting

17. 10 campaigns targeting 1 person (zip code, gender,
workplace, alma mater)
Korolova, “Privacy Violations Using Microtargeted Ads: A Case Study”, PADM
Facebook vs Korolova
Age
21
22
23
…
30
Ad Impressions in a week
0
0
8
…
0

18. 10 campaigns targeting 1 person (zip code, gender,
workplace, alma mater)
Korolova, “Privacy Violations Using Microtargeted Ads: A Case Study”, PADM
Facebook vs Korolova
Interest
A
B
C
…
Z
Ad Impressions in a week
0
0
8
…
0

19. ● Context: Microtargeted Ads
● Takeaway: Attackers can instrument ad campaigns to
identify individual users.
● Two types of attacks:
○ Inference from Impressions
○ Inference from Clicks
Facebook vs Korolova: Recap

20. Attacker's Advantage
Auxiliary information
Enough to succeed on a small fraction of inputs
High dimensionality
Active

21. Items frequently bought together
Bought: A B C D E
Z: Customers Who Bought This Item Also Bought
Calandrino, Kilzer, Narayanan, Felten, Shmatikov, “You Might Also Like: Privacy Risks of Collaborative
Attacking Amazon.com
A C D E

22. Attacker's Advantage
Auxiliary information
Enough to succeed on a small fraction of inputs
High dimensionality
Active
Observant

23. Homer et al., “Resolving individuals contributing trace
amounts of DNA to highly complex mixtures using high-
density SNP genotyping microarrays”, PLoS Genetics,
2008
Genetic data

24. Reference population
Bayesian Analysis

26. “In all mixtures, the identification
of the presence of a person’s
genomic DNA was possible.”

27. Zerhouni, NIH Director:
“As a result, the NIH has removed from
open-access databases the aggregate
results (including P values and genotype
counts) for all the GWAS that had been
available on NIH sites”
… one week later

28. Attacker's Advantage
Auxiliary information
Enough to succeed on a small fraction of inputs
High dimensionality
Active
Observant
Clever

29. Differential Privacy

30. Curator
Defining Privacy

31. 31
CuratorCurator
Your data in
the database
Defining Privacy
Your data in
the database

32. Defining Privacy
32
CuratorCurator
Intuition:
● A member’s privacy is preserved if …
○ “The released result would nearly be the same, whether or
not the user’s information is taken into account”
● An attacker gains very little additional knowledge about any specific member from the
published result
Defining Privacy
Your data in
the database

33. 33
CuratorCurator
Databases D and D′ are neighbors if they differ in one person’s data.
Differential Privacy [DMNS06]: The distribution of the curator’s output M(D)
on database D is (nearly) the same as M(D′).
Differential Privacy
Your data in
the database
Your data in
the database

34. ε-Differential Privacy: The distribution of the curator’s output M(D) on
database D is (nearly) the same as M(D′).
34
CuratorCurator
Parameter ε quantifies
information leakage
∀S: Pr[M(D)∊S] ≤ exp(ε) ∙ Pr[M(D′)∊S].
Differential Privacy
Your data in
the database
Your data in
the database
Dwork, McSherry, Nissim, Smith [TCC 2006]

35. (ε, δ)-Differential Privacy: The distribution of the curator’s output M(D) on
database D is (nearly) the same as M(D′).
35
Parameter ε quantifies
information leakage
Parameter δ allows for
a small probability of
failure
∀S: Pr[M(D)∊S] ≤ exp(ε) ∙ Pr[M(D′)∊S]+δ.
CuratorCurator
Dwork, McSherry, Nissim, Smith [TCC 2006]; Dwork, Kenthapadi, McSherry, Mironov, Naor [ EUROCRYPT 2006]
Your data in
the database
Your data in
the database
Differential Privacy

36. 36
f(D) f(D′)
— bad outcomes
— probability with record x
— probability without record x
“Bad Outcomes” Interpretation

37. ● Prior on databases p
● Observed output O
● Does the database contain record x?
37
Bayesian Interpretation

38. ● Robustness to auxiliary data
● Post-processing:
If M(D) is differentially private, so is f(M(D)).
● Composability:
Run two ε-DP mechanisms. Full interaction is 2ε-DP.
● Group privacy:
Graceful degradation in the presence of
correlated inputs. 38
Differential Privacy

39. Differential Privacy: Takeaway points
• Privacy as a notion of stability of randomized algorithms in
respect to small perturbations in their input
• Worst-case definition
• Robust (to auxiliary data, correlated inputs)
• Composable
• Quantifiable
• Concept of a privacy budget
• Noise injection

40. Case Studies

41. Google’s RAPPOR

42. 44
London, 1854: Broad Street Cholera Outbreak

43. ...Mountain View, 2014

45. Central Model
Curator

46. Local Model

47. Randomized response:
Collecting a sensitive Boolean
Developed in the 1960s for sensitive surveys
“Have you had an abortion?”
- flip a coin, in private
- if coin lands heads, respond “YES”
- if coin lands tails, respond with the truth
Unbiased estimate calculated as: 2 × (fraction of “YES” - ½ )

48. Randomized response:
Collecting a sensitive Boolean
Developed in the 1960s for sensitive surveys
“Have you had an abortion?”
- flip a coin, in private
- if coin lands heads, respond “YES”
flip another coin to respond “YES” or “NO”
- if coin lands tails, respond with the truth
Unbiased estimate calculated as: 2 × (fraction of “YES” - ½ )
Satisfies differential privacy

49. RAPPOR
Erlingsson, Pihur, Korolova. "RAPPOR: Randomized aggregatable privacy-preserving
ordinal response." ACM CCS 2014.

50. RAPPOR: two-level randomized response
Can we do repeated surveys of sensitive attributes?
— Average of randomized responses will reveal a user’s true answer :-(
Solution: Memoize! Re-use the same random answer
— Memoization can hurt privacy too! Long, random bit sequence can
be a unique tracking ID :-(
Solution: Use 2-levels! Randomize the memoized response

51. RAPPOR: two-level randomized response
● Store client value v into bloom filter B using hash functions
● Memoize a Permanent Randomized Response (PRR) B′
● Report an Instantaneous Randomized Response (IRR) S

52. RAPPOR: two-level randomized response
● Store client value v into bloom filter B using hash functions
● Memoize a Permanent Randomized Response (PRR) B′
● Report an Instantaneous Randomized Response (IRR) S
f = ½
q = ¾ , p = ½

53. RAPPOR: Life of a report
Value
Bloom
Filter
PRR
IRR
“www.google.com”

54. Value
Bloom
Filter
PRR
IRR
“www.google.com”
P(1) =
0.25
P(1) =
0.75
RAPPOR: Life of a report

55. Value
Bloom
Filter
PRR
IRR
“www.google.com”
P(1) =
0.50
P(1) =
0.75
RAPPOR: Life of a report

56. Differential privacy of RAPPOR
● Permanent Randomized Response satisfies differential privacy at
● Instantaneous Randomized Response has differential privacy at
= 4 ln(3)
= ln(3)

57. Differential Privacy of RAPPOR:
Measurable privacy bounds
Each report offers differential privacy with
ε = ln(3)
Attacker’s guess goes from 0.1% → 0.3% in the worst case
Differential privacy even if attacker gets all reports (infinite data!!!)
Also… Base Rate Fallacy prevents attackers from finding needles in
haystacks

58. Cohorts
Bloom Filter: 2 bits out of 128 — too many false positives
...
user 0xA0FE91B76:
google.com
cohort 2cohort 1 cohort 128
h2

59. Decoding RAPPOR

60. From Raw Counts to De-noised Counts
True bit counts, with no noise
De-noised RAPPOR reports

61. From De-Noised Count to Distribution
True bit counts, with no noise
De-noised RAPPOR reports
google.com:
yahoo.com:
bing.com:

62. From De-Noised Count to Distribution
Linear Regression:
minX ||B - A X||2
LASSO:
minX (||B - A X||2)2 + λ||X||1
Hybrid:
1. Find support of X via LASSO
2. Solve linear regression to find weights

63. Deploying RAPPOR

64. Coverage

65. Explaining RAPPOR
“Having the cake and eating it too…”
“Seeing the forest without seeing the trees…”

66. Metaphor for RAPPOR

67. Microdata: An Individual’s Report

68. Microdata: An Individual’s Report
Each bit is flipped with
probability
25%

69. Big Picture Remains!

70. Google Chrome Privacy White Paper
https://www.google.com/chrome/browser/privacy/whitepaper.html
Phishing and malware protection
Google Chrome includes an optional feature called "Safe Browsing" to help protect you against phishing and malware attacks. This
helps prevent evil-doers from tricking you into sharing personal information with them (“phishing”) or installing malicious software
on your computer (“malware”). The approach used to accomplish this was designed specifically to protect your privacy and is also
used by other popular browsers.
If you'd rather not send any information to Safe Browsing, you can also turn these features off. Please be aware that Chrome will no
longer be able to protect you from websites that try to steal your information or install harmful software if you disable this feature.
We really don't recommend turning it off.
…
If a URL was indeed dangerous, Chrome reports this anonymously to Google to improve Safe Browsing. The data sent is randomized,
constructed in a manner that ensures differential privacy, permitting only monitoring of aggregate statistics that apply to tens of
thousands of users at minimum. The reports are an instance of Randomized Aggregatable Privacy-Preserving Ordinal Responses,
whose full technical details have been published in a technical report and presented at the 2014 ACM Computer and Communications
Security conference. This means that Google cannot infer which website you have visited from this.

71. Developers’ Uptake

72. RAPPOR:
Lessons Learned

73. Growing Pains
● Transitioning from a research prototype to a real product
● Scalability
● Versioning

74. Communicating Uncertainty

75. Maintaining Candidates List
No missing candidates Three missing candidates
4%
13% 17%

76. RAPPOR Metrics in Chrome
https://chromium.googlesource.com/chromium/src/+log/master/tools/metrics/rappor/rappor.xml

77. Open Source Efforts
https://github.com/google/rappor
- demo you can run with a couple
of shell commands
- client library in several languages
- analysis tool and simulation
- documentation

78. Follow-up
- Bassily, Smith, “Local, Private, Efficient Protocols for Succinct
Histograms,” STOC 2015
- Kairouz, Bonawitz, Ramage, “Discrete Distribution Estimation under
Local Privacy”, https://arxiv.org/abs/1602.07387
- Qin et al., “Heavy Hitter Estimation over Set-Valued Data with Local
Differential Privacy”, CCS 2016

79. Key takeaway points
RAPPOR - locally differentially-private mechanism for reporting of
categorical and string data
● First Internet-scale deployment of differential privacy
● Explainable
● Conservative
● Open-sourced

80. Apple's On-Device Differential
Privacy
Abhradeep Thakurta, UC Santa Cruz

81. Apple WWDC, June 2016

82. References
https://arxiv.org/abs/1709.02753

83. Phablet
Derp
Photobomb
Woot
Phablet
OMG
Woot
Troll
Prepone
Phablet
awwww
dp
Learning from private data
Learn new (and frequent) words typed

84. Learning from private data
Learn frequent emojis typed

85. Apple's On-Device Differential
Privacy: Discovering New Words

86. Roadmap
1. Private frequency estimation with count-min-sketch
2. Private heavy hitters with puzzle piece algorithm
3. Private heavy hitters with tree histogram protocol

87. Private Frequency Oracle

88. Private frequency oracle
Building block for private heavy hitters
𝑑2𝑑1 𝑑 𝑛
All errors within
𝛾 = O( 𝑛 log|𝒮|)
frequency
Words (𝒮)
𝛾
"phablet"
frequency("phablet")

89. Private frequency oracle:
Design constraints
Computational and communication constraints:
Client side:
size of the domain (|S|) and n
Communication to server:
very few bits
Server-side cost for one query:
size of the domain (|S|) and n

90. Private frequency oracle:
Design constraints
Computational and communication constraints:
Client side:
size of the domain (|S|) and n
# characters > 3,000
For 8-character words:
size of the domain |S|=3,000^8
number of clients ~ 1B
Efficiently [BS15] ~ n
Our goal ~ O(log |S|)

91. Private frequency oracle:
Design constraints
Computational and communication constraints:
Client side:
O(log |S|)
Communication to server:
O(1) bits
Server-side cost for one query:
O(log |S|)

92. Private frequency oracle
A starter solution: Randomized response
𝑑
0 1 0
𝑖
1 0 1
𝑖
Protects ε-differential privacy
(with the right bias)
Randomized response: d′

93. 1 0 0
1 1 0
1 0 1
+ With bias
correction
frequency
All domain elements
Error in each estimate:
Θ( 𝑛 log|𝒮|)
Optimal error under privacy
Private frequency oracle
A starter solution: Randomized response

94. Computational and communication constraints:
Client side:
O(|S|)
Communication to server:
O(|S|) bits
Server-side cost for one query:
O(1)
Private frequency oracle
A starter solution: Randomized response
1 0 1
𝑖

95. 𝑑
0 01
0 01
0 01
Hash function: ℎ1
Hash function: ℎ2
Hash function: ℎ 𝑘
Number of hash bins: 𝑛
Computation= 𝑂(log|𝒮|)
𝑘 ≈ log|𝒮|
Private frequency oracle
Non-private count-min sketch [CM05]

96. 0 01
0 01
0 01
0 01
1 00
0 11
1
𝑘
1
+
245
127
9123
2132
𝑛
Reducing server computation
Private frequency oracle
Non-private count-min sketch [CM05]

97. Reducing server computation
1
𝑘
1
Phablet
245
127
9123
2132
𝑛
9146
2212
Frequency estimate:
min (9146, 2212, 2132)
Error in each estimate:
O( 𝑛log|𝒮|)
Server side query cost:
𝑂(log|𝒮|)
𝑘 ≈ log |𝒮|
Private frequency oracle
Non-private count-min sketch [CM05]
"phablet"

98. Private frequency oracle
Private count-min sketch
𝑑
Making client computation differentially private
0 01
0 01
0 01
1 01
1 00
0 00
𝑘𝜖-diff. private, since 𝑘 pieces of information

99. Private frequency oracle
Private count-min sketch
𝑑
Theorem: Sampling ensures 𝜖-differential privacy without hurting accuracy,
rather improves it by a factor of 𝑘
0 01 1 00

100. Private frequency oracle
Private count-min sketch
0 01 +1 +1-1
Hadamard transform
Reducing client communication

101. Private frequency oracle
Private count-min sketch
0 01 +1 +1-1
Hadamard transform
-1 +1
Communication: 𝑂(1) bit
Theorem: Hadamard transform and sampling
do not hurt accuracy
Reducing client communication

102. Private frequency oracle
Private count-min sketch
Error in each estimate:
O( 𝑛log|𝒮|)
Computational and communication constraints:
Client side:
O(log |S|)
Communication to server:
O(1) bits
Server-side cost for one query:
O(log |S|)

103. Roadmap
1. Private frequency estimation with count-min-sketch
2. Private heavy hitters with puzzle piece algorithm
3. Private heavy hitters with tree histogram protocol

104. Private heavy hitters:
Using the frequency oracle
Private frequency oracle
Private count-min sketch
Domain 𝒮
Too many elements in 𝒮 to search.
Element s in S
Frequency(s)
Find all s in S with
frequency > γ

105. Puzzle piece algorithm
(works well in practice, no theoretical guarantees)
[Bassily Nissim Stemmer Thakurta, 2017 and Apple differential privacy team, 2017]

106. Private heavy hitters
Ph ab le t$ Frequency > 𝛾
Each bi-gram frequency > 𝛾
Observation: If a word is frequent, its bigrams are frequent too.

107. Private heavy hitters
Sanitized
bi-grams, and the
complete word
ab
ad
ph
ba
ab
ax
le
ab
ab
Position P1 Position P2 Position P3
le
ab
t$
Position P4
Frequent bi-grams
Natural algorithm: Cartesian product of frequent bi-grams

108. Private heavy hitters
ab
ad
ph
ba
ab
ax
le
ab
ab
Position P1 Position P2 Position P3
le
ab
t$
Position P4
Frequent bi-grams Candidate words
P1 x P2 x P3 x P4
Private frequency oracle
Private count-min sketch
Find frequent
words
Natural algorithm: Cartesian product of frequent bi-grams

109. Private heavy hitters
Candidate words
P1 x P2 x P3 x P4
Private frequency oracle
Find frequent
words
Combinatorial explosion
In practice, all bi-grams are frequent
Natural algorithm: Cartesian product of frequent bi-grams
Private count-min sketch

110. Puzzle piece algorithm
Ph ab le t$
≜
h=Hash(Phablet)
Hash: 𝒮 → 1, … , ℓ
Ph ab le t$h h h h
Privatized
bi-grams tagged
with the hash, and
the complete
word

111. Puzzle piece algorithm: Server side
ab 1
ad 5
Ph 3
ba 4
ab 3
ax 9
le 3
le 7
ab 1
Position P1 Position P2 Position P3
le 1
ab 9
t$ 3
Position P4
Frequent bi-grams tagged with {1, … , ℓ}
Candidate words
P1 x P2 x P3 x P4
Private frequency oracle
Find frequent
words
Combine only matching
bi-grams
Private count-min sketch

112. Roadmap
1. Private frequency estimation with count-min-sketch
2. Private heavy hitters with puzzle piece algorithm
3. Private heavy hitters with tree histogram protocol

113. Tree histogram algorithm
(works well in practice + optimal theoretical guarantees)
[Bassily Nissim Stemmer Thakurta, 2017]

114. Private heavy hitters:
Tree histograms (based on [CM05])
1 0 0
Any string in 𝒮:
log |𝒮| bits
Idea: Construct prefixes of the heavy hitter bit by bit

115. Private heavy hitters:
Tree histograms
0 1

116. Private heavy hitters:
Tree histograms
0 1
Level 1: Frequent prefix of length 1
Use private frequency oracle
If a string is a heavy hitter, its prefixes are too.

117. Private heavy hitters:
Tree histograms
00 01 10 11

118. Private heavy hitters:
Tree histograms
Level 2: Frequent prefix of length two
Idea: Each level has ≈ 𝑛 heavy hitters
00 01 10 11

119. Private heavy hitters:
Tree histograms
Theorem: Finds all heavy hitters with frequency at least
𝑂( 𝑛 log|𝒮|)
Computational and communication constraints:
Client side:
O(log |S|)
Communication to server:
O(1) bits
Server-side computation:
O(n log |S|)

120. Key takeaway points
• Keeping local differential privacy constant:
•One low-noise report is better than many noisy ones
•Weak signal with probability 1 is better than strong signal with small probability
• We can learn the dictionary – at a cost
• Longitudinal privacy remains a challenge

121. NIPS 2017

122. Microsoft: Discretization of continuous variables
"These guarantees are particularly strong when user’s behavior remains
approximately the same, varies slowly, or varies around a small number of
values over the course of data collection."

123. Microsoft's deployment
"Our mechanisms have been deployed by Microsoft
across millions of devices ... to protect users’ privacy
while collecting application usage statistics."
B. Ding, J. Kulkarni, S. Yekhanin, NIPS 2017
More details: today, 1pm-5pm
T12: Privacy at Scale: Local Differential Privacy in
Practice, G. Cormode, T. Kulkarni, N. Li, T. Wang

124. LinkedIn Salary

125. Outline
• LinkedIn Salary Overview
• Challenges: Privacy, Modeling
• System Design & Architecture
• Privacy vs. Modeling Tradeoffs

126. LinkedIn Salary (launched in Nov, 2016)

127. Salary Collection Flow via Email Targeting

128. Current Reach (August 2018)
• A few million responses out of several millions of members targeted
• Targeted via emails since early 2016
• Countries: US, CA, UK, DE, IN, …
• Insights available for a large fraction of US monthly active users

129. Data Privacy Challenges
• Minimize the risk of inferring any one individual’s compensation data
• Protection against data breach
• No single point of failure
Achieved by a combination of
techniques: encryption, access control,
, aggregation,
thresholding
K. Kenthapadi, A. Chudhary, and S.
Ambler, LinkedIn Salary: A System
for Secure Collection and
Presentation of Structured
Compensation Insights to Job
Seekers, IEEE PAC 2017
(arxiv.org/abs/1705.06976)

130. Modeling Challenges
• Evaluation
• Modeling on de-identified data
• Robustness and stability
• Outlier detection
X. Chen, Y. Liu, L. Zhang, and K.
Kenthapadi, How LinkedIn
Economic Graph Bonds
Information and Product:
Applications in LinkedIn Salary,
KDD 2018
(arxiv.org/abs/1806.09063)
K. Kenthapadi, S. Ambler,
L. Zhang, and D. Agarwal,
Bringing salary transparency to
the world: Computing robust
compensation insights via
LinkedIn Salary, CIKM 2017
(arxiv.org/abs/1703.09845)

131. Problem Statement
•How do we design LinkedIn Salary system taking into
account the unique privacy and security challenges,
while addressing the product requirements?

132. Differential Privacy? [Dwork et al, 2006]
• Rich privacy literature (Adam-Worthmann, Samarati-Sweeney, Agrawal-Srikant, …,
Kenthapadi et al, Machanavajjhala et al, Li et al, Dwork et al)
• Limitation of anonymization techniques (as discussed in the first part)
• Worst case sensitivity of quantiles to any one user’s compensation data is
large
•  Large noise to be added, depriving reliability/usefulness
• Need compensation insights on a continual basis
• Theoretical work on applying differential privacy under continual observations
• No practical implementations / applications
• Local differential privacy / Randomized response based approaches (Google’s RAPPOR; Apple’s
iOS differential privacy; Microsoft’s telemetry collection) not applicable

133. Title Region
$$
User Exp
Designer
SF Bay
Area 100K
User Exp
Designer
SF Bay
Area 115K
... ...
...
Title Region
$$
User Exp
Designer
SF Bay
Area 100K
De-identification Example
Title Region Company Industry Years of
exp
Degree FoS Skills
$$
User Exp
Designer
SF Bay
Area
Google Internet 12 BS Interactive
Media
UX,
Graphics,
...
100K
Title Region Industry
$$
User Exp
Designer
SF Bay
Area
Internet
100K
Title Region Years of
exp $$
User Exp
Designer
SF Bay
Area
10+
100K
Title Region Company Years of
exp $$
User Exp
Designer
SF Bay
Area
Google 10+
100K
#data
points >
threshold?
Yes ⇒ Copy to
Hadoop (HDFS) Note: Original submission stored as encrypted objects.

134. System
Architecture

135. Collection
&
Storage

136. Collection & Storage
• Allow members to submit their compensation info
• Extract member attributes
• E.g., canonical job title, company, region, by invoking LinkedIn standardization services
• Securely store member attributes & compensation data

137. De-identification
&
Grouping

138. De-identification & Grouping
• Approach inspired by k-Anonymity [Samarati-Sweeney]
• “Cohort” or “Slice”
• Defined by a combination of attributes
• E.g, “User experience designers in SF Bay Area”
• Contains aggregated compensation entries from corresponding individuals
• No user name, id or any attributes other than those that define the cohort
• A cohort available for offline processing only if it has at least k entries
• Apply LinkedIn standardization software (free-form attribute  canonical version)
before grouping
• Analogous to the generalization step in k-Anonymity

139. De-identification & Grouping
• Slicing service
• Access member attribute info &
submission identifiers (no
compensation data)
• Generate slices & track #
submissions for each slice
• Preparation service
• Fetch compensation data (using
submission identifiers), associate
with the slice data, copy to HDFS

140. Insights
&
Modeling

141. Insights & Modeling
• Salary insight service
• Check whether the member is
eligible
• Give-to-get model
• If yes, show the insights
• Offline workflow
• Consume de-identified HDFS
dataset
• Compute robust compensation
insights
• Outlier detection
• Bayesian smoothing/inference
• Populate the insight key-value
stores

142. Security
Mechanisms

143. Security
Mechanisms
• Encryption of
member attributes
& compensation
data using different
sets of keys
• Separation of
processing
• Limiting access to
the keys

144. Security
Mechanisms
• Key rotation
• No single point of
failure
• Infra security

145. Preventing Timestamp Join based Attacks
• Inference attack by joining these on timestamp
• De-identified compensation data
• Page view logs (when a member accessed compensation collection web interface)
•  Not desirable to retain the exact timestamp
• Perturb by adding random delay (say, up to 48 hours)
• Modification based on k-Anonymity
• Generalization using a hierarchy of timestamps
• But, need to be incremental
•  Process entries within a cohort in batches of size k
• Generalize to a common timestamp
• Make additional data available only in such incremental batches

146. Privacy vs Modeling Tradeoffs
• LinkedIn Salary system deployed in production for ~2.5 years
• Study tradeoffs between privacy guarantees (‘k’) and data available for
computing insights
• Dataset: Compensation submission history from 1.5M LinkedIn members
• Amount of data available vs. minimum threshold, k
• Effect of processing entries in batches of size, k

147. Amount of
data
available vs.
threshold, k

148. Percent of
data available
vs. batch size,
k

149. Median delay
due to
batching vs.
batch size, k

150. Key takeaway points
• LinkedIn Salary: a new internet application, with
unique privacy/modeling challenges
• Privacy vs. Modeling Tradeoffs
• Potential directions
• Privacy-preserving machine learning models in a practical setting
[e.g., Chaudhuri et al, JMLR 2011; Papernot et al, ICLR 2017]
• Provably private submission of compensation entries?

151. Beyond Randomized Response

152. Beyond Randomized Response
• LDP + Machine Learning:
• "Is interaction necessary for distributed private learning?"
Smith, Thakurta, Upadhyay, S&P 2017
• Federated Learning
• Encode-Shuffle-Analyze architecture
"Prochlo: Strong Privacy for Analytics in the Crowd"
Bittau et al., SOSP 2017
• Amplification by Shuffling

153. LDP + Machine Learning
Interactivity as a major implementation constraint
...
parallel

154. LDP + Machine Learning
Interactivity as a major implementation constraint
...
sequential

155. "Is interaction necessary for distributed private
learning?" [STU 2017]
• Single parameter learning (e.g., median):
• Maximal accuracy with full parallelism
• Multi-parameter learning:
• Polylog number of iterations
• Lower bounds

156. Federated Learning
"Practical secure aggregation for privacy-preserving machine learning"
Bonawitz, Ivanov, Kreuter, Marcedone, McMahan, Patel, Ramage, Segal,
Seth, ACM CCS 2017

157. PROCHLO:
Strong Privacy for Analytics in the Crowd
Bittau, Erlingsson, Maniatis, Mironov, Raghunathan,
Lie, Rudominer, Kode, Tinnes, Seefeld
SOSP 2017

158. The ESA Architecture and Its Prochlo realization
E
A
E
E
S
Σ
ESA: Encode, Shuffle, Analyze (ESA)
Prochlo: A hardened ESA realization using Intel's SGX + crypto

159. E
A
E
E
S
Σ
S S...
A
Σ
Σ
Σ
Local DP
Unlinkability
Randomized Thresholding Central DP

160. Key takeaway points
• Notion of differential privacy is a principled foundation for privacy-
preserving data analyses
• Local differential privacy is a powerful technique appropriate for
Internet-scale telemetry
• Other techniques (thresholding, shuffling) can be combined with
differentially private algorithms or be used in isolation.

161. References
Differential privacy:
review "A Firm Foundation For Private Data Analysis", C. ACM 2011
by Dwork
book "The Algorithmic Foundations of Differential Privacy"
by Dwork and Roth

162. References
Google's RAPPOR:
paper "RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal
Response", ACM CCS 2014, Erlingsson, Pihur, Korolova
blog
Apple's implementation:
article "Learning with Privacy at Scale", Apple ML J., Dec 2017
paper "Practical Locally Private Heavy Hitters", NIPS 2017,
by Bassily, Nissim, Stemmer, Thakurta
paper "Privacy Loss in Apple's Implementation of Differential Privacy on
MacOS 10.12" by Tang, Korolova, Bai, Wang, Wang
LinkedIn Salary:
paper "LinkedIn Salary: A System for Secure Collection and Presentation of
Structured Compensation Insights to Job Seekers", IEEE PAC 2017,
Kenthapadi, Chudhary, Ambler
blog

163. Thanks & Questions
• Tutorial website: https://sites.google.com/view/kdd2018privacytutorial
• Feedback most welcome 
• kkenthapadi@linkedin.com, mironov@google.com
• Related KDD’18 tutorial today afternoon (1pm-5pm):
T12: Privacy at Scale: Local Differential Privacy in Practice, G. Cormode,
T. Kulkarni, N. Li, T. Wang