5. Challenges:
To protect each participant’s private data set and
intermediate results.
The computation/ communication cost introduced to
each participant shall be affordable.
For collaborative training, training data is arbitrarily
partitioned.
5
INTRODUCTION(Contd..)
6. Provides privacy preservation for multiparty .
Collaborative BPN network learning over arbitrarily
partitioned data.
Guarantees privacy and efficiency.
Support multiparty secure scalar product.
Allow decryption of arbitrary large messages.
6
CONTRIBUTONS
7. System Model:
Trusted authority.
The participating parties ( data owner).
The cloud servers ( or cloud).
Security Model:
7
MODELS AND ASSUMPTIONS
8. Arbitrarily Partitioned Data
Z parties (Z > 2 ) : Ps , 1 ≤ s ≤ Z.
Database D with N rows : {DB1,DB2, ….. DBN}.
Each row DBv ,1 ≤ v ≤ N has m attributes {xv
1 , xv
2 , xv
3 …..
xv
m}.
DBv = DBv
1 U DBv
2 U DBv
3 U ….. U DBv
z .
Each DBv, Ps has ts
v attributes.
8
TECHNIQUE PRELIMINARIES
9. BACK –PROPOGATION NEURAL NETWORK LEARNING
9
TECHNIQUE PRELIMINARIES(Contd..)
10. BGN Homomorphic Encryption
Operations on plaintexts to be performed on their
respective cipher texts.
Public-key “doubly homomorphic” encryption
scheme(called “BGN” for short).
One multiplication and unlimited number of additions.
Given ciphertexts C(m1) , C(m2) and C(m^1), C(m^2 ), one
can compute C(m1 m^1 + m2m^2) without knowing the
plaintext.
10
TECHNIQUE PRELIMINARIES(Contd..)
11. PROBLEM STATEMENT
3 layer (a-b-c configuration) neural network .
N samples for learning data set .
Arbitrary partitioned into Z( Z≥2) subsets.
SCHEME OVERVIEW
Each party encrypt her/his input data set.
Participants upload the encrypted data to cloud.
Cloud servers perform the operations.
Secret sharing algorithm.
11
PROPOSED SCHEME
25. ACCURACY ANALYSIS
Accuracy loss in approximation of activation function.
Maclaurin series used – accuracy can be adjusted by
modifying number of series terms.
25
PERFORMANCE EVALUATION(Contd..)
26. Secure and practical multiparty BPN network learning.
Cost independent of number of parties.
Scalable efficient and secure.
26
CONCLUSION
27. 1) N. Schlitter A Protocol for Privacy Preserving Neural
Network Learning on Horizontal Partitioned Data, Proc.
Privacy Statistics in Databases (PSD ’08), Sept. 2008
2) T. Chen and S. Zhong, Privacy-Preserving
Backpropagation Neural Network Learning,IEEE Trans.
Neural Network, vol. 20, no. 10, Oct. 2000,pp. 1554-1564
3) A. Bansal, T. Chen, and S. Zhong, Privacy Preserving
Back-Propagation Neural Network Learning over
Arbitrarily Parti-tioned Data,Neural Computing
Applications,vol. 20, no. 1, Feb. 2011, pp. 143-150,
4) D. Boneh, E.-J. Goh, and K. Nissim, Evaluating 2-DNF
Formulas on Ciphertexts,Proc. Second Int’l Conf. Theory
of Cryptography (TCC ’05), pp. 325-341, 2005.
27
REFERENCES