Author : Md Abdul Aziz 1
Date of Publication :15th March 2017
Abstract: Now a days, data is growing rapidly in different fields like trade reports, social networks, hospital etc and all these data maintained in cloud. To extract the data from cloud, we can use data mining concepts. Classifying these data, the time series data has become a topic of great interest in data mining. In case only one person is dealing with the data there are many algorithms to classify. In this paper, we are classifying the data when it is with two or more persons and no one wants to share their data, even though we can do classification of the data securely. There are two basic data partitioning models, they are horizontal partitioning and vertical partitioning. We are proposing novel time series classification algorithm for classification of vertically partitioned data. Our approach is on distance based novel classification of the time series data. To classify securely we use the concept of secure multi-party computation protocols. By using secure multiparty computation protocols, we are proposing secure square root sum and secure sum protocols for classifying data securely. In these two approaches secure square root sum gives the best result.
Reference :
-
- IEEE paper by P.Ravi Kumar department of CSE fast classification of time series data using novel classification in 2010.
- Tingting Chen, Sheng Zhong Privacy Preserving Back Propagation-chen- zhong-2009 using Secure multiplication on time series data
- D. Agarwal and C.C Agarwal On the design and qualification of Privacy Preserving Data mining algorithms.
- William Stallings Cryptography and Network Security Principles and practice 5 th edition.
- Jiawei Han and Micheline Kamber - Data Mining. Concepts and Techniques 2 nd Edition.