A Practical Differentially Private Random Decision Tree Classifier

Jagannathan, Geetha; Pillaipakkamnatt, Krishnan; Wright, Rebecca N.

doi:10.1109/icdmw.2009.93

Cited by 127 publications

(135 citation statements)

References 14 publications

(18 reference statements)

Supporting

Mentioning

133

Contrasting

Order By: Relevance

“…With the success of differentially private algorithms for tabular data [16,8,1,25,26,13,18,5], there has been interest in developing differentially private mechanisms for other domains, such as social networks [12]. For social networks, examples include [17,27,30].…”

Section: Related Workmentioning

confidence: 99%

“…Query answering algorithms that satisfy differential privacy must produce noisy query answers such that the distribution of query answers changes very little with the addition, deletion, or modification of any tuple. Recent work has shown that the resulting query answers can enable very accurate privacy-preserving statistical analyses of sensitive datasets [25,13,5,18]. There are two flavors of differential privacy, which we call unbounded and bounded.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

No free lunch in data privacy

Kifer

Machanavajjhala

2011

Proceedings of the 2011 ACM SIGMOD International Conference on Management of Data

488

390

View full text Add to dashboard Cite

Differential privacy is a powerful tool for providing privacypreserving noisy query answers over statistical databases. It guarantees that the distribution of noisy query answers changes very little with the addition or deletion of any tuple. It is frequently accompanied by popularized claims that it provides privacy without any assumptions about the data and that it protects against attackers who know all but one record.In this paper we critically analyze the privacy protections offered by differential privacy.First, we use a no-free-lunch theorem, which defines nonprivacy as a game, to argue that it is not possible to provide privacy and utility without making assumptions about how the data are generated. Then we explain where assumptions are needed. We argue that privacy of an individual is preserved when it is possible to limit the inference of an attacker about the participation of the individual in the data generating process. This is different from limiting the inference about the presence of a tuple (for example, Bob's participation in a social network may cause edges to form between pairs of his friends, so that it affects more than just the tuple labeled as "Bob"). The definition of evidence of participation, in turn, depends on how the data are generated -this is how assumptions enter the picture. We explain these ideas using examples from social network research as well as tabular data for which deterministic statistics have been previously released. In both cases the notion of participation varies, the use of differential privacy can lead to privacy breaches, and differential privacy does not always adequately limit inference about participation.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

No free lunch in data privacy

Kifer

Machanavajjhala

2011

Proceedings of the 2011 ACM SIGMOD International Conference on Management of Data

488

390

View full text Add to dashboard Cite

show abstract

“…The merit of building a random tree is the efficiency of training and minimal memory requirements. To create a random decision tree, Random Tree algorithm uses only one pass over the data [17,19].…”

Section: Random Treementioning

confidence: 99%

Predicting Student Performance in Higher Education Institutions Using Decision Tree Analysis

Hamoud¹,

Hashim²,

Awadh³

2018

IJIMAI

131

View full text Add to dashboard Cite

The overall success of educational institutions can be measured by the success of its students. Providing factors that increase success rate and reduce the failure of students is profoundly helpful to educational organizations. Data mining is the best solution to finding hidden patterns and giving suggestions that enhance the performance of students. This paper presents a model based on decision tree algorithms and suggests the best algorithm based on performance. Three built classifiers (J48, Random Tree and REPTree) were used in this model with the questionnaires filled in by students. The survey consists of 60 questions that cover the fields, such as health, social activity, relationships, and academic performance, most related to and affect the performance of students. A total of 161 questionnaires were collected. The Weka 3.8 tool was used to construct this model. Finally, the J48 algorithm was considered as the best algorithm based on its performance compared with the Random Tree and RepTree algorithms.

show abstract

“…Although it is possible to extend binary classification algorithms to multi-class using techniques like one-vsall, it is much more expensive to do so as compared to a naturally multi-class classification algorithm. Jagannathan, et al [11] present a differentially private random decision tree learning algorithm which can be applied to multi-class classification. Their approach involves perturbing leaf nodes using the sensitivity method, and they do not provide theoretical analysis of excess risk of the perturbed classifier.…”

Section: Related Workmentioning

confidence: 99%

Large Margin Multiclass Gaussian Classification with Differential Privacy

Pathak

Raj

2011

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. As increasing amounts of sensitive personal information is aggregated into data repositories, it has become important to develop mechanisms for processing the data without revealing information about individual data instances. The differential privacy model provides a framework for the development and theoretical analysis of such mechanisms. In this paper, we propose an algorithm for learning a discriminatively trained multi-class Gaussian classifier that satisfies differential privacy using a large margin loss function with a perturbed regularization term. We present a theoretical upper bound on the excess risk of the classifier introduced by the perturbation.

show abstract

A Practical Differentially Private Random Decision Tree Classifier

Cited by 127 publications

References 14 publications

No free lunch in data privacy

No free lunch in data privacy

Predicting Student Performance in Higher Education Institutions Using Decision Tree Analysis

Large Margin Multiclass Gaussian Classification with Differential Privacy

Contact Info

Product

Resources

About