Towards Attack-Resilient Geometric Data Perturbation
نویسندگان
چکیده
Data perturbation is a popular technique for privacypreserving data mining. The major challenge of data perturbation is balancing privacy protection and data quality, which are normally considered as a pair of contradictive factors. We propose that selectively preserving only the task/model specific information in perturbation would improve the balance. Geometric data perturbation, consisting of random rotation perturbation, random translation perturbation, and noise addition, aims at preserving the important geometric properties of a multidimensional dataset, while providing better privacy guarantee for data classification modeling. The preliminary study has shown that random geometric perturbation can well preserve model accuracy for several popular classification models, including kernel methods, linear classifiers, and SVM classifiers, while it also revealed some security concerns to random geometric perturbation. In this paper, we address some potential attacks to random geometric perturbation and design several methods to reduce the threat of these attacks. Experimental study shows that the enhanced geometric perturbation can provide satisfactory privacy guarantee while still well preserving model accuracy for the discussed data classification models.
منابع مشابه
A Distributed Deterministic and Resilient Replication Attack Detection Protocol in Wireless Sensor Networks
To detect replica nodes in a wireless sensor network, we propose a distributed, deterministic and resilient (DDR) replica detection protocol developed from a witness node based strategy. In DDR, while a location claim message of each node is sent towards the designated verification location in the network, the consistency of the messages is verified at intermediate nodes en route to its final d...
متن کاملPh.D. Prospectus Towards Practical and Resilient Geometric Routing for Wireless Sensor Networks
متن کامل
An Effective Attack-Resilient Kalman Filter-Based Approach for Dynamic State Estimation of Synchronous Machine
Kalman filtering has been widely considered for dynamic state estimation in smart grids. Despite its unique merits, the Kalman Filter (KF)-based dynamic state estimation can be undesirably influenced by cyber adversarial attacks that can potentially be launched against the communication links in the Cyber-Physical System (CPS). To enhance the security of KF-based state estimation, in this paper...
متن کاملResilient Configuration of Distribution System versus False Data Injection Attacks Against State Estimation
State estimation is used in power systems to estimate grid variables based on meter measurements. Unfortunately, power grids are vulnerable to cyber-attacks. Reducing cyber-attacks against state estimation is necessary to ensure power system safe and reliable operation. False data injection (FDI) is a type of cyber-attack that tampers with measurements. This paper proposes network reconfigurati...
متن کاملA Random Rotation Perturbation Approach to Privacy Preserving Data Classification
This paper presents a random rotation perturbation approach for privacy preserving data classification. Concretely, we identify the importance of classification-specific information with respect to the loss of information factor, and present a random rotation perturbation framework for privacy preserving data classification. Our approach has two unique characteristics. First, we identify that m...
متن کامل