模式分类过程涉及到对原始训练样本的学习，容易导致用户隐私的泄露。为了避免模式分类过程中的隐私泄露，同时又不影响模式分类算法的性能，提出一种基于主成分分析（PCA）的模式分类隐私保护算法。该算法利用PCA 提取原始训练数据的主成分，并将原始训练样本集合转化为主成分的新样本集合，然后利用新样本集合进行分类学习。选用Adult 数据集和KDDCUP 99 数据集进行仿真实验，并采用正确率和召回率进行性能评价，结果表明，该隐私保护算法通过PCA 提取原始数据特征属性的主成分，可避免原始属性的泄露，同时PCA 在一定程度上可实现去噪，从而使分类器的分类性能优于原始数据集的分类性能。与已有算法比较，该隐私保护算法具有更好的模式分类精度和隐私保护性能。
The pattern classification process involves the learning from the original training samples, which easily leads to privacy disclosure. In order to avoid the leaks of privacy in the pattern classification process and not to affect the performance of the algorithm, this paper proposes a pattern classification privacy preserve algorithm based on the primary component analysis (PCA). This algorithm extracts the principal component of the original training data and converts the original training samples to new samples corresponding to the primary components. Then, a classification model is trained on the new samples. Experiments are carried out on the Adult data set and the KDD CUP 99 data set, and the precision and recall indexes are used to evaluate the proposed algorithm. It is shown that this algorithm can avoid the leakage of the original attributes through extracting the principal components of the feature attributes about the raw data. PCA can achieve de-noising to some extent, so that the classification performance on the classifier is better than that on the original data set. Therefore, compared with the existing algorithms, this algorithm has better pattern classification accuracy and privacy preserve performance.
[1] Han J, Kamber M, Pei J. Data mining: Concepts and techniques[M]. CA,San Mateo: Morgan kaufmann, 2006.
[2] Sweeney L. k-anonymity: A model for protecting privacy[J]. International Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 2002, 10(5): 557-570.
[3] Machanavajjhala A, Kifer D, Gehrke J, et al. L-diversity: Privacy beyond k-anonymity[J]. ACM Transactions on Knowledge Discovery from Data, 2007(1): 3.
[4] 田秀霞, 王晓玲, 高明, 等. 数据库服务——安全与隐私保护[J]. 软件 学报, 2010, 21(5): 991-1006. Tian Xiuxia, Wang Xiaoling, Gao Ming, et al. Database as a service? security and privacy preserving[J]. Journal of Software, 2010, 21(5): 991-1006.
[5] Yang J, Yu X, Xie Z Q, et al. A novel virtual sample generation method based on Gaussian distribution[J]. Knowledge-Based Systems, 2011, 24(6): 740-748.
[6] 戴群, 陈松灿, 王喆. 一个基于自组织特征映射网络的混合神经网络 结构[J]. 软件学报, 2009, 20(5): 1329-1336. Dai Qun, Chen Songcan, Wang Zhe. Hybrid neural network architecture based on self-organizing feature maps[J]. Journal of Software, 2009, 20(5): 1329-1336.
[7] 杨静, 辛宇, 谢志强. 面向物联网传感器事件监测的双向反馈系统[J]. 计算机学报, 2013, 36(3): 506-520. Yang Jing, Xin Yu, Xie Zhiqiang. A bi-feedback system of wireless sensor network event detection in the internet of things[J]. Chinese Journal of Computers, 2013, 36(3): 506-520.
[8] Cortes C, Vapnik V. Support-vector networks[J]. Machine learning, 1995, 20(3): 273-297.
[9] 曾志强, 高济. 基于向量集约简的精简支持向量机[J]. 软件学报, 2007, 18(11): 2719-2727. Zeng Zhiqiang, Gao Ji. Simplified support vector machine based on reduced vector set method[J]. Journal of Software, 2007, 18(11): 2719-2727.
[10] 顾彬, 郑关胜, 王建东. 增量和减量式标准支持向量机的分析[J]. 软 件学报, 2013, 24(7): 1601-1613. Gu Bin, Zheng Guansheng, Wang Jiandong. Analysis for incremental and decremental standard support vector machine[J]. Journal of Software, 2013, 24(7): 1601-1613.
[11] Quinlan J R. C4.5: Programs for machine learning[M]. San Mateo, CA: Morgan Kaufmann, 1993.
[12] Zhou Z H, Jiang Y. NeC4.5: Neural ensemble based C4.5[J]. Knowledge and Data Engineering, IEEE Transactions on, 2004, 16(6): 770-773.
[13] Breiman L, Friedman J, Stone C J, et al. Classification and regression trees[M]. Florida: CRC Press, 1984.
[14] Agrawal R, Srikant R. Privacy-preserving data mining[J]. ACM Sigmod Record, 2000, 29(2): 439-450.
[15] Kargupta H, Datta S, Wang Q, et al. On the privacy preserving properties of random data perturbation techniques[C]//Data Mining, 2003. Third IEEE International Conference on. New York: IEEE, 2003: 99-106.
[16] Bapna S, Gangopadhyay A. A wavelet-based approach to preserve privacy for classification mining[J]. Decision Sciences, 2006, 37(4): 623-642.
[17] 胡文军, 王士同. 隐私保护的SVM 快速分类方法[J] . 电子学报, 2012, 40(2): 280-286. Hu Wenjun, Wang Shitong. Fast classification approach of support vector machine with privacy preservation[J]. Acta Electronica Sinica, 2012, 40(2): 280-286.
[18] Xiao X, Tao Y. Personalized privacy preservation[C]//Proceedings of the 2006 ACM SIGMOD International Conference on Management of data. Chicago: ACM, 2006: 229-240.
[19] Duda R O, Hart P E, Stork D G. Pattern classification[M]. New York: John Wiley & Sons, 2012.
[20] Kohavi R. A study of cross-validation and bootstrap for accuracy estimation and model selection[C]// IJCAI'95 Proceedings of the 14th International Joint Conference on Artificial Intelligence, 1995, 2: 1137-1145.
[21] 董春曦. 支持向量机及其在入侵检测中的应用研究[D]. 西安: 西安 电子科技大学, 2004. Dong Chunxi. Study of support vector machines and its application in intrusion detection systems[D]. Xi'an: Xidian University, 2004.
