A Genetic Approach to Multivariate Microaggregation for Database Privacy
- 1 April 2007
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 1, 180-185
Abstract
Microaggregation is a technique used to protect privacy in databases and location-based services. We propose a new hybrid technique for multivariate microaggregation. Our technique combines a heuristic yielding fixed-size groups and a genetic algorithm yielding variable-sized groups. Fixed-size heuristics are fast and able to deal with large data sets, but they sometimes are far from optimal in terms of the information loss inflicted. On the other hand, the genetic algorithm obtains very good results (i.e. optimal or near optimal), but it can only cope with very small data sets. Our technique leverages the advantages of both types of heuristics and avoids their shortcomings. First, it partitions the data set into a number of groups by using a fixed-size heuristic. Then. it optimizes the partitions by means of the genetic algorithm. As an outcome of this mixture of heuristics, we obtain a technique that improves the results of the fixed-size heuristic in large data sets.Keywords
This publication has 13 references indexed in Scilit:
- Efficient multivariate data-oriented microaggregationThe VLDB Journal, 2006
- A Fixed Structure Learning Automaton Micro-aggregation Technique for Secure Statistical DatabasesLecture Notes in Computer Science, 2006
- Ordinal, Continuous and Heterogeneous k-Anonymity Through MicroaggregationData Mining and Knowledge Discovery, 2005
- A polynomial algorithm for optimal univariate microaggregationIEEE Transactions on Knowledge and Data Engineering, 2003
- EXACT AND APPROXIMATE METHODS FOR DATA DIRECTED MICROAGGREGATION IN ONE OR MORE DIMENSIONSInternational Journal of Uncertainty, Fuzziness and Knowledge-Based Systems, 2002
- Practical data-oriented microaggregation for statistical disclosure controlIEEE Transactions on Knowledge and Data Engineering, 2002
- Minimum Sum of Squares Clustering in a Low Dimensional SpaceJournal of Classification, 1998
- An Algorithm for Euclidean Sum of Squares ClassificationBiometrics, 1977
- A Method for Cluster AnalysisBiometrics, 1965
- Hierarchical Grouping to Optimize an Objective FunctionJournal of the American Statistical Association, 1963