Practical Secure Decision Tree Learning in a Teletreatment Application

Hoogh, Sebastiaan de; Schoenmakers, Berry; Chen, Ping; Akker, Harm op den

doi:10.1007/978-3-662-45472-5_12

Cited by 30 publications

(63 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a criterion we choose to minimize the Gini index, as also used in the CART algorithm, since it only requires a few secure multiplications to compute. 2 For discrete attributes, we follow [13] and compute the Gini index by securely counting the number of elements in the dataset that satisfy certain equations using an indicator-vector representation, as described in Section 4. For continuous attributes, the situation is more complicated since the equations are not based on equality (=), but rather a less-than-or-equal (≤) predicate.…”

Section: Overview Of Our Techniquesmentioning

confidence: 99%

See 1 more Smart Citation

Secure training of decision trees with continuous attributes

Abspoel

Escudero

Volgushev

2020

Proceedings on Privacy Enhancing Technologies

View full text Add to dashboard Cite

We apply multiparty computation (MPC) techniques to show, given a database that is secret-shared among multiple mutually distrustful parties, how the parties may obliviously construct a decision tree based on the secret data. We consider data with continuous attributes (i.e., coming from a large domain), and develop a secure version of a learning algorithm similar to the C4.5 or CART algorithms. Previous MPC-based work only focused on decision tree learning with discrete attributes (De Hoogh et al. 2014). Our starting point is to apply an existing generic MPC protocol to a standard decision tree learning algorithm, which we then optimize in several ways. We exploit the fact that even if we allow the data to have continuous values, which a priori might require fixed or floating point representations, the output of the tree learning algorithm only depends on the relative ordering of the data. By obliviously sorting the data we reduce the number of comparisons needed per node to O(N log2N) from the naive O(N2), where N is the number of training records in the dataset, thus making the algorithm feasible for larger datasets. This does however introduce a problem when duplicate values occur in the dataset, but we manage to overcome this problem with a relatively cheap subprotocol. We show a procedure to convert a sorting network into a permutation network of smaller complexity, resulting in a round complexity of O(log N) per layer in the tree. We implement our algorithm in the MP-SPDZ framework and benchmark our implementation for both passive and active three-party computation using arithmetic modulo 264. We apply our implementation to a large scale medical dataset of ≈ 290 000 rows using random forests, and thus demonstrate practical feasibility of using MPC for privacy-preserving machine learning based on decision trees for large datasets.

show abstract

Section: Overview Of Our Techniquesmentioning

confidence: 99%

“…This issue was addressed in [13], where an extension of ID3 to the secure setting was given using Shamir secret-sharing and allowing arbitrary initial partitioning of the data. However, their protocol does not allow for continuous attributes, which is an important feature of decision trees with respect to other machine learning models.…”

Section: Related Workmentioning

confidence: 99%

Secure training of decision trees with continuous attributes

Abspoel

Escudero

Volgushev

2020

Proceedings on Privacy Enhancing Technologies

View full text Add to dashboard Cite

show abstract

“…SMC based Machine Learning. A significant body of work in PPML with SMC has focused on the problem of privacy-preserving training of machine learning models (see, e.g., [12], [13], [14], [15], [16] and references therein). Privacy-preserving protocols for predicting with trained ML classifiers -hereafter also referred to as "scoring" -has received far less attention.…”

Section: Related Workmentioning

confidence: 99%

“…Bost et al [17] introduced privacypreserving protocols for hyperplane-based, Naive Bayes and DT classifiers, Wu et al [18] for DTs and RFs, David et al [19] for hyperplane-based and Naive Bayes classifiers, and De Cock et al [20] for DTs and hyperplane-based classifiers. De Hoogh et al [14] had also previously presented a protocol for privacy-preserving scoring of DTs with categorical attributes. The protocol for privacy-preserving scoring of DTs of De Cock et al [20] cannot be directly used as a building block to obtain random forests and boosted decision trees.…”

Section: Related Workmentioning

confidence: 99%

Privacy-Preserving Scoring of Tree Ensembles: A Novel Framework for AI in Healthcare

Fritchman

Saminathan

Dowsley

et al. 2018

2018 IEEE International Conference on Big Data (Big Data)

View full text Add to dashboard Cite

Machine Learning (ML) techniques now impact a wide variety of domains. Highly regulated industries such as healthcare and finance have stringent compliance and data governance policies around data sharing. Advances in secure multiparty computation (SMC) for privacy-preserving machine learning (PPML) can help transform these regulated industries by allowing ML computations over encrypted data with personally identifiable information (PII). Yet very little of SMC-based PPML has been put into practice so far. In this paper we present the very first framework for privacypreserving classification of tree ensembles with application in healthcare. We first describe the underlying cryptographic protocols that enable a healthcare organization to send encrypted data securely to a ML scoring service and obtain encrypted class labels without the scoring service actually seeing that input in the clear. We then describe the deployment challenges we solved to integrate these protocols in a cloud based scalable risk-prediction platform with multiple ML models for healthcare AI. Included are system internals, and evaluations of our deployment for supporting physicians to drive better clinical outcomes in an accurate, scalable, and provably secure manner. To the best of our knowledge, this is the first such applied framework with SMC-based privacypreserving machine learning for healthcare.

show abstract

“…In a practical application, the cold start classifier should be trained, and regularly updated by some third party service provider, meaning that possibly privacy sensitive data needs to be send from the user to this third party. A possible solution to this issue is presented by (de Hoogh et al 2014) who have used a portion of the data used for the Kairos system to develop and evaluate algorithms for privacy-preserving data mining, effectively allowing predictions to be made based on encrypted data.…”

Section: Outstanding Issuesmentioning

confidence: 99%

Smart tailoring of real-time physical activity coaching systems

Akker¹

Self Cite

View full text Add to dashboard Cite

Practical Secure Decision Tree Learning in a Teletreatment Application

Cited by 30 publications

References 20 publications

Secure training of decision trees with continuous attributes

Secure training of decision trees with continuous attributes

Privacy-Preserving Scoring of Tree Ensembles: A Novel Framework for AI in Healthcare

Smart tailoring of real-time physical activity coaching systems

Contact Info

Product

Resources

About