Summary
|
For Full-Text PDF, please login, if you are a member of IEICE, or go to Pay Per View on menu list, if you are a nonmember of IEICE. |
|
Privacy-Preserving Decision Tree Learning with Boolean Target Class Hiroaki KIKUCHI Kouichi ITOH Mebae USHIDA Hiroshi TSUDA Yuji YAMAOKA Publication IEICE TRANSACTIONS on Fundamentals of Electronics, Communications and Computer Sciences Vol.E98-A No.11 pp.2291-2300 Publication Date: 2015/11/01 Online ISSN: 1745-1337 DOI: 10.1587/transfun.E98.A.2291 Type of Manuscript: PAPER Category: Cryptography and Information Security Keyword: privacy, decision tree learning, Boolean query, piecewise linear function, Full Text: PDF(1.3MB)>>
Summary: This paper studies a privacy-preserving decision tree learning protocol (PPDT) for vertically partitioned datasets. In vertically partitioned datasets, a single class (target) attribute is shared by both parities or carefully treated by either party in existing studies. The proposed scheme allows both parties to have independent class attributes in a secure way and to combine multiple class attributes in arbitrary boolean function, which gives parties some flexibility in data-mining. Our proposed PPDT protocol reduces the CPU-intensive computation of logarithms by approximating with a piecewise linear function defined by light-weight fundamental operations of addition and constant multiplication so that information gain for attributes can be evaluated in a secure function evaluation scheme. Using the UCI Machine Learning dataset and a synthesized dataset, the proposed protocol is evaluated in terms of its accuracy and the sizes of trees*. | ||||||||||
| ||||||||||
