Scribd
Upload
64 views

Privacy Preserving Data Mining

This document summarizes a paper on privacy-preserving data mining. It discusses how two parties can jointly compute a data mining algorithm like decision tree learning on their private databases without revealing unnecessary private information. It presents an efficient protocol for privately computing the ID3 decision tree algorithm based on secret sharing and oblivious transfer. The parties learn only the output tree structure and not each other's private inputs or intermediate values. This allows meaningful data mining without privacy breaches.

Uploaded by

Prasanthi Prasu
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as RTF, PDF, TXT or read online on Scribd
64 views

Privacy Preserving Data Mining

This document summarizes a paper on privacy-preserving data mining. It discusses how two parties can jointly compute a data mining algorithm like decision tree learning on their private databases without revealing unnecessary private information. It presents an efficient protocol for privately computing the ID3 decision tree algorithm based on secret sharing and oblivious transfer. The parties learn only the output tree structure and not each other's private inputs or intermediate values. This allows meaningful data mining without privacy breaches.

Uploaded by

Prasanthi Prasu
Copyright
© Attribution Non-Commercial (BY-NC)
Available Formats
Download as RTF, PDF, TXT or read online on Scribd
You are on page 1/ 80

Privacy Preserving Data Mining

Yehuda Lindell Benny Pinkas

Presenter: Justin Brickell

Mining Joint Databases


Parties P1 and P2 own databases D1 and D2 f is a data mining algorithm

Compute f(D1 D2) without revealing unnecessary information

Unnecessary Information
Intuitively, the protocol should function as if a trusted third party computed the output
P1 D1 f(D1 D2) D2 f(D1 D2) P2

TTP

Simulation
Let msg(P2) be P2s messages If S1 can simulate msg(P2) to P1 given only P1s input and the protocol output, then msg(P2) must not contain unnecessary information (and viceversa)

S1(D1,f(D1,D2)) =C msg(P2)

More Simulation Details


The simulator S1 can also recover r1, the internal coin tosses ofP1 Can extend to allow distinctf1(x,y) and f2(x,y)
- Complicates the definition

- Not necessary for data mining applications

The Semi-Honest Model


A malicious adversary can alter his input
- f( D2) = f(D2) !

A semi-honest adversary
- adheres to protocol

- tries to learn extra information from the message transcript

General Secure Two Party Computation


Any algorithm can be made private (in the semi-honest model)
- Yaos Protocol

So, why write this paper?


- Yaos Protocol is inefficient

- This paper privately computes aparticular algorithm more efficiently

Yaos Protocol (Basically)


Convert the algorithm to a circuit P1 hard codes his input into the circuit P1 transforms each gate so that it takes garbled inputs to garbled outputs

Using 1-out-of-2 oblivious transfer, P1 sends P2 garbled versions of his inputs

Garbled Wire Values


P1 assigns to each wire i two random values (Wi0,Wi1)
- Long enough to seed a pseudo-random function F

P1 assigns to each wire i a random permutation over {0,1}, i : bi ci

Wibi,ci is the garbled value of wire i

Garbled Gates
Gate g computes bk = g(bi,bj) Garbled gate is a table Tg computing Wibi,ci Wjbj,cj Wkbk,ck
- Tg has four entries:

- ci,cj: Wkg(bi,bj),ck F [Wibi](cj) F [Wjbj](ci)

Yaos Protocol
P1 sends
- P2s garbled input bits (1-out-of-2) - Tg tables - Table from garbled output values to output bits

P2 can compute output values, butP1s input and intermediate values appear random

Cost of circuit with n inputs and m gates


Communication: m gate tables
- 4m length of pseudo-random output

Computation: n oblivious transfers


- Typically much more expensive than them pseudo-random function applications

Too expensive for data mining

Classification by Decision Tree Learning


A classic machine learning / data mining problem Develop rules for when a transaction belongs to a class based on its attribute values Smaller decision trees are better

ID3 is one particular algorithm

A Database
Outlook Sunny Sunny Overcast Rain Rain Rain Overcast Sunny Sunny Rain Sunny Overcast Overcast Temp Humidity Hot High Hot High Mild High Mild High Cool Normal Cool Normal Cool Normal Mild High Cool Normal Mild Normal Mild Normal Mild High Hot Normal Wind Play Tennis Weak No Strong No Weak Yes Weak Yes Weak Yes Strong No Strong Yes Weak No Weak Yes Weak Yes Strong Yes Strong Yes Weak Yes

Rain

Mild

High

Strong

No

and its Decision Tree


Outlook Sunny Humidity High Normal Yes Rain Overcast Yes Wind Strong No Weak Yes

The ID3 Algorithm: Definitions


R: The set of attributes
- Outlook, Temperature, Humidity, Wind

C: the class attribute


- Play Tennis

T: the set of transactions

- The 14 database entries

The ID3 Algorithm


ID3(R,C,T) If R is empty, return a leaf-node with the most common class value in T If all transactions in T have the same class value c, return the leaf-nodec Otherwise,
- Determine the attribute A that best classifies T - Create a tree node labeled A, recur to compute child trees

- edge ai goes to tree ID3(R - {A},C,T(ai))

The Best Predicting Attribute


Entropy! l
HC T

( )=
(

T (c )
i

log

T ci

( )
T

i= i

T
m

HC T | A

)=
j= i

T aj T

( )

Gain(A) =def HC(T) - HC(T|A)

( )

HC T aj

( )

Find A with maximum gain

Why can we do better than Yao?


Normally, private protocols must hide intermediate values In this protocol, the assignment of attributes to nodes ispart of the output and may be revealed
- H values are not revealed, just the identity of the attribute with greatest gain

This allows genuine recursion

How do we do it?
Rather than maximize gain, minimize
- HC(T|A) =def HC(T|A)|T|ln 2

This has the simple formula


HC T | A

)=
m l j= 1i= i

T aj,ci

( ) ln(

) ()( )
T aj
j= 1

+
m

ln

Terms have form (v1+v2)ln(v1+v2)

T aj,ci

( ) T(a )
j

- P1 knows v1, P2 knows v2

Private x ln x
Input: P1s value v1, P2s value v2 Auxiliary Input: A large field F Output: P1 obtains w1 F, P2 obtains w2 F
- w1 + w2 (v1 + v2)ln(v1+v2)

- w1 and w2 are uniformly distributed in F

Private x ln x: some intuition


Compute shares of x and ln x, then privately multiply Shares of ln x are actually shares of n and where x = 2n(1+)
- -1/2 1/2

Uses Taylor expansions

Using the x ln x protocol


For every attribute A, every attributevalue aj A, and every class ci C
- wA,1(aj), wA,2(aj), wA,1(aj,ci), wA,2(aj,ci) - wA,1(aj) + wA,2(aj) |T(aj)|ln(|T(aj)| - wA,1(aj,ci) + wA,2(aj,ci)

|T(aj,ci)|ln(|T(aj,ci)|

Shares of Relative Entropy


P1 and P2 can locally compute shares SA,1 + SA,2 HC(T|A)

Now, use the Yao protocol to find theA with minimum Relative Entropy!

A Technical Detail
The logarithms are only approximate
- ID3 algorithm - Doesnt distinguish relative entropies within

Complexity for each node


For |R| attributes, m attribute values, and l class values
- x ln x protocol is invoked O(m l |R|) times - Each requires O(log|T|) oblivious transfers - And bandwidth O(k log|T| |S|) bits
k depends logarithmically on

Depends only logarithmically on |T|

Only k|S| worse that non-private distributed ID3

Conclusion
Private computation of ID3(D1 D2) is made feasible Using Yaos protocol directly would be impractical

Questions?

You might also like